git/t/t0021-conversion.sh

705 lines
20 KiB
Bash
Raw Normal View History

#!/bin/sh
test_description='blob conversion via gitattributes'
. ./test-lib.sh
TEST_ROOT="$PWD"
PATH=$TEST_ROOT:$PATH
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
write_script <<\EOF "$TEST_ROOT/rot13.sh"
tr \
'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ' \
'nopqrstuvwxyzabcdefghijklmNOPQRSTUVWXYZABCDEFGHIJKLM'
EOF
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
write_script rot13-filter.pl "$PERL_PATH" \
<"$TEST_DIRECTORY"/t0021/rot13-filter.pl
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
generate_random_characters () {
LEN=$1
NAME=$2
test-genrandom some-seed $LEN |
perl -pe "s/./chr((ord($&) % 26) + ord('a'))/sge" >"$TEST_ROOT/$NAME"
}
file_size () {
perl -e 'print -s $ARGV[0]' "$1"
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
}
filter_git () {
rm -f rot13-filter.log &&
git "$@"
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
}
# Compare two files and ensure that `clean` and `smudge` respectively are
# called at least once if specified in the `expect` file. The actual
# invocation count is not relevant because their number can vary.
# c.f. http://public-inbox.org/git/xmqqshv18i8i.fsf@gitster.mtv.corp.google.com/
test_cmp_count () {
expect=$1
actual=$2
for FILE in "$expect" "$actual"
do
sort "$FILE" | uniq -c |
sed -e "s/^ *[0-9][0-9]*[ ]*IN: /x IN: /" >"$FILE.tmp" &&
mv "$FILE.tmp" "$FILE" || return
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
done &&
test_cmp "$expect" "$actual"
}
# Compare two files but exclude all `clean` invocations because Git can
# call `clean` zero or more times.
# c.f. http://public-inbox.org/git/xmqqshv18i8i.fsf@gitster.mtv.corp.google.com/
test_cmp_exclude_clean () {
expect=$1
actual=$2
for FILE in "$expect" "$actual"
do
grep -v "IN: clean" "$FILE" >"$FILE.tmp" &&
mv "$FILE.tmp" "$FILE"
done &&
test_cmp "$expect" "$actual"
}
# Check that the contents of two files are equal and that their rot13 version
# is equal to the committed content.
test_cmp_committed_rot13 () {
test_cmp "$1" "$2" &&
rot13.sh <"$1" >expected &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
git cat-file blob :"$2" >actual &&
test_cmp expected actual
}
test_expect_success setup '
git config filter.rot13.smudge ./rot13.sh &&
git config filter.rot13.clean ./rot13.sh &&
{
echo "*.t filter=rot13"
echo "*.i ident"
} >.gitattributes &&
{
echo a b c d e f g h i j k l m
echo n o p q r s t u v w x y z
echo '\''$Id$'\''
} >test &&
cat test >test.t &&
cat test >test.o &&
cat test >test.i &&
git add test test.t test.i &&
rm -f test test.t test.i &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
git checkout -- test test.t test.i &&
echo "content-test2" >test2.o &&
echo "content-test3 - filename with special characters" >"test3 '\''sq'\'',\$x=.o"
'
script='s/^\$Id: \([0-9a-f]*\) \$/\1/p'
test_expect_success check '
test_cmp test.o test &&
test_cmp test.o test.t &&
# ident should be stripped in the repository
git diff --raw --exit-code :test :test.i &&
id=$(git rev-parse --verify :test) &&
embedded=$(sed -ne "$script" test.i) &&
test "z$id" = "z$embedded" &&
git cat-file blob :test.t >test.r &&
./rot13.sh <test.o >test.t &&
test_cmp test.r test.t
'
# If an expanded ident ever gets into the repository, we want to make sure that
# it is collapsed before being expanded again on checkout
test_expect_success expanded_in_repo '
{
echo "File with expanded keywords"
echo "\$Id\$"
echo "\$Id:\$"
echo "\$Id: 0000000000000000000000000000000000000000 \$"
echo "\$Id: NoSpaceAtEnd\$"
echo "\$Id:NoSpaceAtFront \$"
echo "\$Id:NoSpaceAtEitherEnd\$"
echo "\$Id: NoTerminatingSymbol"
echo "\$Id: Foreign Commit With Spaces \$"
} >expanded-keywords.0 &&
{
cat expanded-keywords.0 &&
printf "\$Id: NoTerminatingSymbolAtEOF"
} >expanded-keywords &&
cat expanded-keywords >expanded-keywords-crlf &&
git add expanded-keywords expanded-keywords-crlf &&
git commit -m "File with keywords expanded" &&
id=$(git rev-parse --verify :expanded-keywords) &&
{
echo "File with expanded keywords"
echo "\$Id: $id \$"
echo "\$Id: $id \$"
echo "\$Id: $id \$"
echo "\$Id: $id \$"
echo "\$Id: $id \$"
echo "\$Id: $id \$"
echo "\$Id: NoTerminatingSymbol"
echo "\$Id: Foreign Commit With Spaces \$"
} >expected-output.0 &&
{
cat expected-output.0 &&
printf "\$Id: NoTerminatingSymbolAtEOF"
} >expected-output &&
{
append_cr <expected-output.0 &&
printf "\$Id: NoTerminatingSymbolAtEOF"
} >expected-output-crlf &&
{
echo "expanded-keywords ident"
echo "expanded-keywords-crlf ident text eol=crlf"
} >>.gitattributes &&
rm -f expanded-keywords expanded-keywords-crlf &&
git checkout -- expanded-keywords &&
test_cmp expanded-keywords expected-output &&
git checkout -- expanded-keywords-crlf &&
test_cmp expanded-keywords-crlf expected-output-crlf
'
# The use of %f in a filter definition is expanded to the path to
# the filename being smudged or cleaned. It must be shell escaped.
# First, set up some interesting file names and pet them in
# .gitattributes.
test_expect_success 'filter shell-escaped filenames' '
cat >argc.sh <<-EOF &&
#!$SHELL_PATH
cat >/dev/null
echo argc: \$# "\$@"
EOF
normal=name-no-magic &&
special="name with '\''sq'\'' and \$x" &&
echo some test text >"$normal" &&
echo some test text >"$special" &&
git add "$normal" "$special" &&
git commit -q -m "add files" &&
echo "name* filter=argc" >.gitattributes &&
# delete the files and check them out again, using a smudge filter
# that will count the args and echo the command-line back to us
test_config filter.argc.smudge "sh ./argc.sh %f" &&
rm "$normal" "$special" &&
git checkout -- "$normal" "$special" &&
# make sure argc.sh counted the right number of args
echo "argc: 1 $normal" >expect &&
test_cmp expect "$normal" &&
echo "argc: 1 $special" >expect &&
test_cmp expect "$special" &&
# do the same thing, but with more args in the filter expression
test_config filter.argc.smudge "sh ./argc.sh %f --my-extra-arg" &&
rm "$normal" "$special" &&
git checkout -- "$normal" "$special" &&
# make sure argc.sh counted the right number of args
echo "argc: 2 $normal --my-extra-arg" >expect &&
test_cmp expect "$normal" &&
echo "argc: 2 $special --my-extra-arg" >expect &&
test_cmp expect "$special" &&
:
'
convert: stream from fd to required clean filter to reduce used address space The data is streamed to the filter process anyway. Better avoid mapping the file if possible. This is especially useful if a clean filter reduces the size, for example if it computes a sha1 for binary data, like git media. The file size that the previous implementation could handle was limited by the available address space; large files for example could not be handled with (32-bit) msysgit. The new implementation can filter files of any size as long as the filter output is small enough. The new code path is only taken if the filter is required. The filter consumes data directly from the fd. If it fails, the original data is not immediately available. The condition can easily be handled as a fatal error, which is expected for a required filter anyway. If the filter was not required, the condition would need to be handled in a different way, like seeking to 0 and reading the data. But this would require more restructuring of the code and is probably not worth it. The obvious approach of falling back to reading all data would not help achieving the main purpose of this patch, which is to handle large files with limited address space. If reading all data is an option, we can simply take the old code path right away and mmap the entire file. The environment variable GIT_MMAP_LIMIT, which has been introduced in a previous commit is used to test that the expected code path is taken. A related test that exercises required filters is modified to verify that the data actually has been modified on its way from the file system to the object store. Signed-off-by: Steffen Prohaska <prohaska@zib.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2014-08-26 23:23:25 +08:00
test_expect_success 'required filter should filter data' '
test_config filter.required.smudge ./rot13.sh &&
test_config filter.required.clean ./rot13.sh &&
test_config filter.required.required true &&
echo "*.r filter=required" >.gitattributes &&
convert: stream from fd to required clean filter to reduce used address space The data is streamed to the filter process anyway. Better avoid mapping the file if possible. This is especially useful if a clean filter reduces the size, for example if it computes a sha1 for binary data, like git media. The file size that the previous implementation could handle was limited by the available address space; large files for example could not be handled with (32-bit) msysgit. The new implementation can filter files of any size as long as the filter output is small enough. The new code path is only taken if the filter is required. The filter consumes data directly from the fd. If it fails, the original data is not immediately available. The condition can easily be handled as a fatal error, which is expected for a required filter anyway. If the filter was not required, the condition would need to be handled in a different way, like seeking to 0 and reading the data. But this would require more restructuring of the code and is probably not worth it. The obvious approach of falling back to reading all data would not help achieving the main purpose of this patch, which is to handle large files with limited address space. If reading all data is an option, we can simply take the old code path right away and mmap the entire file. The environment variable GIT_MMAP_LIMIT, which has been introduced in a previous commit is used to test that the expected code path is taken. A related test that exercises required filters is modified to verify that the data actually has been modified on its way from the file system to the object store. Signed-off-by: Steffen Prohaska <prohaska@zib.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2014-08-26 23:23:25 +08:00
cat test.o >test.r &&
git add test.r &&
convert: stream from fd to required clean filter to reduce used address space The data is streamed to the filter process anyway. Better avoid mapping the file if possible. This is especially useful if a clean filter reduces the size, for example if it computes a sha1 for binary data, like git media. The file size that the previous implementation could handle was limited by the available address space; large files for example could not be handled with (32-bit) msysgit. The new implementation can filter files of any size as long as the filter output is small enough. The new code path is only taken if the filter is required. The filter consumes data directly from the fd. If it fails, the original data is not immediately available. The condition can easily be handled as a fatal error, which is expected for a required filter anyway. If the filter was not required, the condition would need to be handled in a different way, like seeking to 0 and reading the data. But this would require more restructuring of the code and is probably not worth it. The obvious approach of falling back to reading all data would not help achieving the main purpose of this patch, which is to handle large files with limited address space. If reading all data is an option, we can simply take the old code path right away and mmap the entire file. The environment variable GIT_MMAP_LIMIT, which has been introduced in a previous commit is used to test that the expected code path is taken. A related test that exercises required filters is modified to verify that the data actually has been modified on its way from the file system to the object store. Signed-off-by: Steffen Prohaska <prohaska@zib.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2014-08-26 23:23:25 +08:00
rm -f test.r &&
convert: stream from fd to required clean filter to reduce used address space The data is streamed to the filter process anyway. Better avoid mapping the file if possible. This is especially useful if a clean filter reduces the size, for example if it computes a sha1 for binary data, like git media. The file size that the previous implementation could handle was limited by the available address space; large files for example could not be handled with (32-bit) msysgit. The new implementation can filter files of any size as long as the filter output is small enough. The new code path is only taken if the filter is required. The filter consumes data directly from the fd. If it fails, the original data is not immediately available. The condition can easily be handled as a fatal error, which is expected for a required filter anyway. If the filter was not required, the condition would need to be handled in a different way, like seeking to 0 and reading the data. But this would require more restructuring of the code and is probably not worth it. The obvious approach of falling back to reading all data would not help achieving the main purpose of this patch, which is to handle large files with limited address space. If reading all data is an option, we can simply take the old code path right away and mmap the entire file. The environment variable GIT_MMAP_LIMIT, which has been introduced in a previous commit is used to test that the expected code path is taken. A related test that exercises required filters is modified to verify that the data actually has been modified on its way from the file system to the object store. Signed-off-by: Steffen Prohaska <prohaska@zib.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2014-08-26 23:23:25 +08:00
git checkout -- test.r &&
test_cmp test.o test.r &&
convert: stream from fd to required clean filter to reduce used address space The data is streamed to the filter process anyway. Better avoid mapping the file if possible. This is especially useful if a clean filter reduces the size, for example if it computes a sha1 for binary data, like git media. The file size that the previous implementation could handle was limited by the available address space; large files for example could not be handled with (32-bit) msysgit. The new implementation can filter files of any size as long as the filter output is small enough. The new code path is only taken if the filter is required. The filter consumes data directly from the fd. If it fails, the original data is not immediately available. The condition can easily be handled as a fatal error, which is expected for a required filter anyway. If the filter was not required, the condition would need to be handled in a different way, like seeking to 0 and reading the data. But this would require more restructuring of the code and is probably not worth it. The obvious approach of falling back to reading all data would not help achieving the main purpose of this patch, which is to handle large files with limited address space. If reading all data is an option, we can simply take the old code path right away and mmap the entire file. The environment variable GIT_MMAP_LIMIT, which has been introduced in a previous commit is used to test that the expected code path is taken. A related test that exercises required filters is modified to verify that the data actually has been modified on its way from the file system to the object store. Signed-off-by: Steffen Prohaska <prohaska@zib.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2014-08-26 23:23:25 +08:00
./rot13.sh <test.o >expected &&
git cat-file blob :test.r >actual &&
test_cmp expected actual
'
test_expect_success 'required filter smudge failure' '
test_config filter.failsmudge.smudge false &&
test_config filter.failsmudge.clean cat &&
test_config filter.failsmudge.required true &&
echo "*.fs filter=failsmudge" >.gitattributes &&
echo test >test.fs &&
git add test.fs &&
rm -f test.fs &&
test_must_fail git checkout -- test.fs
'
test_expect_success 'required filter clean failure' '
test_config filter.failclean.smudge cat &&
test_config filter.failclean.clean false &&
test_config filter.failclean.required true &&
echo "*.fc filter=failclean" >.gitattributes &&
echo test >test.fc &&
test_must_fail git add test.fc
'
convert: stream from fd to required clean filter to reduce used address space The data is streamed to the filter process anyway. Better avoid mapping the file if possible. This is especially useful if a clean filter reduces the size, for example if it computes a sha1 for binary data, like git media. The file size that the previous implementation could handle was limited by the available address space; large files for example could not be handled with (32-bit) msysgit. The new implementation can filter files of any size as long as the filter output is small enough. The new code path is only taken if the filter is required. The filter consumes data directly from the fd. If it fails, the original data is not immediately available. The condition can easily be handled as a fatal error, which is expected for a required filter anyway. If the filter was not required, the condition would need to be handled in a different way, like seeking to 0 and reading the data. But this would require more restructuring of the code and is probably not worth it. The obvious approach of falling back to reading all data would not help achieving the main purpose of this patch, which is to handle large files with limited address space. If reading all data is an option, we can simply take the old code path right away and mmap the entire file. The environment variable GIT_MMAP_LIMIT, which has been introduced in a previous commit is used to test that the expected code path is taken. A related test that exercises required filters is modified to verify that the data actually has been modified on its way from the file system to the object store. Signed-off-by: Steffen Prohaska <prohaska@zib.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2014-08-26 23:23:25 +08:00
test_expect_success 'filtering large input to small output should use little memory' '
test_config filter.devnull.clean "cat >/dev/null" &&
test_config filter.devnull.required true &&
convert: stream from fd to required clean filter to reduce used address space The data is streamed to the filter process anyway. Better avoid mapping the file if possible. This is especially useful if a clean filter reduces the size, for example if it computes a sha1 for binary data, like git media. The file size that the previous implementation could handle was limited by the available address space; large files for example could not be handled with (32-bit) msysgit. The new implementation can filter files of any size as long as the filter output is small enough. The new code path is only taken if the filter is required. The filter consumes data directly from the fd. If it fails, the original data is not immediately available. The condition can easily be handled as a fatal error, which is expected for a required filter anyway. If the filter was not required, the condition would need to be handled in a different way, like seeking to 0 and reading the data. But this would require more restructuring of the code and is probably not worth it. The obvious approach of falling back to reading all data would not help achieving the main purpose of this patch, which is to handle large files with limited address space. If reading all data is an option, we can simply take the old code path right away and mmap the entire file. The environment variable GIT_MMAP_LIMIT, which has been introduced in a previous commit is used to test that the expected code path is taken. A related test that exercises required filters is modified to verify that the data actually has been modified on its way from the file system to the object store. Signed-off-by: Steffen Prohaska <prohaska@zib.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2014-08-26 23:23:25 +08:00
for i in $(test_seq 1 30); do printf "%1048576d" 1; done >30MB &&
echo "30MB filter=devnull" >.gitattributes &&
GIT_MMAP_LIMIT=1m GIT_ALLOC_LIMIT=1m git add 30MB
'
test_expect_success 'filter that does not read is fine' '
test-genrandom foo $((128 * 1024 + 1)) >big &&
echo "big filter=epipe" >.gitattributes &&
test_config filter.epipe.clean "echo xyzzy" &&
git add big &&
git cat-file blob :big >actual &&
echo xyzzy >expect &&
test_cmp expect actual
'
xread, xwrite: limit size of IO to 8MB Checking out 2GB or more through an external filter (see test) fails on Mac OS X 10.8.4 (12E55) for a 64-bit executable with: error: read from external filter cat failed error: cannot feed the input to external filter cat error: cat died of signal 13 error: external filter cat failed 141 error: external filter cat failed The reason is that read() immediately returns with EINVAL when asked to read more than 2GB. According to POSIX [1], if the value of nbyte passed to read() is greater than SSIZE_MAX, the result is implementation-defined. The write function has the same restriction [2]. Since OS X still supports running 32-bit executables, the 32-bit limit (SSIZE_MAX = INT_MAX = 2GB - 1) seems to be also imposed on 64-bit executables under certain conditions. For write, the problem has been addressed earlier [6c642a]. Address the problem for read() and write() differently, by limiting size of IO chunks unconditionally on all platforms in xread() and xwrite(). Large chunks only cause problems, like causing latencies when killing the process, even if OS X was not buggy. Doing IO in reasonably sized smaller chunks should have no negative impact on performance. The compat wrapper clipped_write() introduced earlier [6c642a] is not needed anymore. It will be reverted in a separate commit. The new test catches read and write problems. Note that 'git add' exits with 0 even if it prints filtering errors to stderr. The test, therefore, checks stderr. 'git add' should probably be changed (sometime in another commit) to exit with nonzero if filtering fails. The test could then be changed to use test_must_fail. Thanks to the following people for suggestions and testing: Johannes Sixt <j6t@kdbg.org> John Keeping <john@keeping.me.uk> Jonathan Nieder <jrnieder@gmail.com> Kyle J. McKay <mackyle@gmail.com> Linus Torvalds <torvalds@linux-foundation.org> Torsten Bögershausen <tboegi@web.de> [1] http://pubs.opengroup.org/onlinepubs/009695399/functions/read.html [2] http://pubs.opengroup.org/onlinepubs/009695399/functions/write.html [6c642a] commit 6c642a878688adf46b226903858b53e2d31ac5c3 compate/clipped-write.c: large write(2) fails on Mac OS X/XNU Signed-off-by: Steffen Prohaska <prohaska@zib.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-08-20 14:43:54 +08:00
test_expect_success EXPENSIVE 'filter large file' '
test_config filter.largefile.smudge cat &&
test_config filter.largefile.clean cat &&
xread, xwrite: limit size of IO to 8MB Checking out 2GB or more through an external filter (see test) fails on Mac OS X 10.8.4 (12E55) for a 64-bit executable with: error: read from external filter cat failed error: cannot feed the input to external filter cat error: cat died of signal 13 error: external filter cat failed 141 error: external filter cat failed The reason is that read() immediately returns with EINVAL when asked to read more than 2GB. According to POSIX [1], if the value of nbyte passed to read() is greater than SSIZE_MAX, the result is implementation-defined. The write function has the same restriction [2]. Since OS X still supports running 32-bit executables, the 32-bit limit (SSIZE_MAX = INT_MAX = 2GB - 1) seems to be also imposed on 64-bit executables under certain conditions. For write, the problem has been addressed earlier [6c642a]. Address the problem for read() and write() differently, by limiting size of IO chunks unconditionally on all platforms in xread() and xwrite(). Large chunks only cause problems, like causing latencies when killing the process, even if OS X was not buggy. Doing IO in reasonably sized smaller chunks should have no negative impact on performance. The compat wrapper clipped_write() introduced earlier [6c642a] is not needed anymore. It will be reverted in a separate commit. The new test catches read and write problems. Note that 'git add' exits with 0 even if it prints filtering errors to stderr. The test, therefore, checks stderr. 'git add' should probably be changed (sometime in another commit) to exit with nonzero if filtering fails. The test could then be changed to use test_must_fail. Thanks to the following people for suggestions and testing: Johannes Sixt <j6t@kdbg.org> John Keeping <john@keeping.me.uk> Jonathan Nieder <jrnieder@gmail.com> Kyle J. McKay <mackyle@gmail.com> Linus Torvalds <torvalds@linux-foundation.org> Torsten Bögershausen <tboegi@web.de> [1] http://pubs.opengroup.org/onlinepubs/009695399/functions/read.html [2] http://pubs.opengroup.org/onlinepubs/009695399/functions/write.html [6c642a] commit 6c642a878688adf46b226903858b53e2d31ac5c3 compate/clipped-write.c: large write(2) fails on Mac OS X/XNU Signed-off-by: Steffen Prohaska <prohaska@zib.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-08-20 14:43:54 +08:00
for i in $(test_seq 1 2048); do printf "%1048576d" 1; done >2GB &&
echo "2GB filter=largefile" >.gitattributes &&
git add 2GB 2>err &&
test_must_be_empty err &&
xread, xwrite: limit size of IO to 8MB Checking out 2GB or more through an external filter (see test) fails on Mac OS X 10.8.4 (12E55) for a 64-bit executable with: error: read from external filter cat failed error: cannot feed the input to external filter cat error: cat died of signal 13 error: external filter cat failed 141 error: external filter cat failed The reason is that read() immediately returns with EINVAL when asked to read more than 2GB. According to POSIX [1], if the value of nbyte passed to read() is greater than SSIZE_MAX, the result is implementation-defined. The write function has the same restriction [2]. Since OS X still supports running 32-bit executables, the 32-bit limit (SSIZE_MAX = INT_MAX = 2GB - 1) seems to be also imposed on 64-bit executables under certain conditions. For write, the problem has been addressed earlier [6c642a]. Address the problem for read() and write() differently, by limiting size of IO chunks unconditionally on all platforms in xread() and xwrite(). Large chunks only cause problems, like causing latencies when killing the process, even if OS X was not buggy. Doing IO in reasonably sized smaller chunks should have no negative impact on performance. The compat wrapper clipped_write() introduced earlier [6c642a] is not needed anymore. It will be reverted in a separate commit. The new test catches read and write problems. Note that 'git add' exits with 0 even if it prints filtering errors to stderr. The test, therefore, checks stderr. 'git add' should probably be changed (sometime in another commit) to exit with nonzero if filtering fails. The test could then be changed to use test_must_fail. Thanks to the following people for suggestions and testing: Johannes Sixt <j6t@kdbg.org> John Keeping <john@keeping.me.uk> Jonathan Nieder <jrnieder@gmail.com> Kyle J. McKay <mackyle@gmail.com> Linus Torvalds <torvalds@linux-foundation.org> Torsten Bögershausen <tboegi@web.de> [1] http://pubs.opengroup.org/onlinepubs/009695399/functions/read.html [2] http://pubs.opengroup.org/onlinepubs/009695399/functions/write.html [6c642a] commit 6c642a878688adf46b226903858b53e2d31ac5c3 compate/clipped-write.c: large write(2) fails on Mac OS X/XNU Signed-off-by: Steffen Prohaska <prohaska@zib.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-08-20 14:43:54 +08:00
rm -f 2GB &&
git checkout -- 2GB 2>err &&
test_must_be_empty err
xread, xwrite: limit size of IO to 8MB Checking out 2GB or more through an external filter (see test) fails on Mac OS X 10.8.4 (12E55) for a 64-bit executable with: error: read from external filter cat failed error: cannot feed the input to external filter cat error: cat died of signal 13 error: external filter cat failed 141 error: external filter cat failed The reason is that read() immediately returns with EINVAL when asked to read more than 2GB. According to POSIX [1], if the value of nbyte passed to read() is greater than SSIZE_MAX, the result is implementation-defined. The write function has the same restriction [2]. Since OS X still supports running 32-bit executables, the 32-bit limit (SSIZE_MAX = INT_MAX = 2GB - 1) seems to be also imposed on 64-bit executables under certain conditions. For write, the problem has been addressed earlier [6c642a]. Address the problem for read() and write() differently, by limiting size of IO chunks unconditionally on all platforms in xread() and xwrite(). Large chunks only cause problems, like causing latencies when killing the process, even if OS X was not buggy. Doing IO in reasonably sized smaller chunks should have no negative impact on performance. The compat wrapper clipped_write() introduced earlier [6c642a] is not needed anymore. It will be reverted in a separate commit. The new test catches read and write problems. Note that 'git add' exits with 0 even if it prints filtering errors to stderr. The test, therefore, checks stderr. 'git add' should probably be changed (sometime in another commit) to exit with nonzero if filtering fails. The test could then be changed to use test_must_fail. Thanks to the following people for suggestions and testing: Johannes Sixt <j6t@kdbg.org> John Keeping <john@keeping.me.uk> Jonathan Nieder <jrnieder@gmail.com> Kyle J. McKay <mackyle@gmail.com> Linus Torvalds <torvalds@linux-foundation.org> Torsten Bögershausen <tboegi@web.de> [1] http://pubs.opengroup.org/onlinepubs/009695399/functions/read.html [2] http://pubs.opengroup.org/onlinepubs/009695399/functions/write.html [6c642a] commit 6c642a878688adf46b226903858b53e2d31ac5c3 compate/clipped-write.c: large write(2) fails on Mac OS X/XNU Signed-off-by: Steffen Prohaska <prohaska@zib.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-08-20 14:43:54 +08:00
'
test_expect_success "filter: clean empty file" '
test_config filter.in-repo-header.clean "echo cleaned && cat" &&
test_config filter.in-repo-header.smudge "sed 1d" &&
echo "empty-in-worktree filter=in-repo-header" >>.gitattributes &&
>empty-in-worktree &&
echo cleaned >expected &&
git add empty-in-worktree &&
git show :empty-in-worktree >actual &&
test_cmp expected actual
'
test_expect_success "filter: smudge empty file" '
test_config filter.empty-in-repo.clean "cat >/dev/null" &&
test_config filter.empty-in-repo.smudge "echo smudged && cat" &&
echo "empty-in-repo filter=empty-in-repo" >>.gitattributes &&
echo dead data walking >empty-in-repo &&
git add empty-in-repo &&
echo smudged >expected &&
git checkout-index --prefix=filtered- empty-in-repo &&
test_cmp expected filtered-empty-in-repo
'
test_expect_success 'disable filter with empty override' '
test_config_global filter.disable.smudge false &&
test_config_global filter.disable.clean false &&
test_config filter.disable.smudge false &&
test_config filter.disable.clean false &&
echo "*.disable filter=disable" >.gitattributes &&
echo test >test.disable &&
git -c filter.disable.clean= add test.disable 2>err &&
test_must_be_empty err &&
rm -f test.disable &&
git -c filter.disable.smudge= checkout -- test.disable 2>err &&
test_must_be_empty err
'
test_expect_success 'diff does not reuse worktree files that need cleaning' '
test_config filter.counter.clean "echo . >>count; sed s/^/clean:/" &&
echo "file filter=counter" >.gitattributes &&
test_commit one file &&
test_commit two file &&
>count &&
git diff-tree -p HEAD &&
test_line_count = 0 count
'
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
test_expect_success PERL 'required process filter should filter data' '
test_config_global filter.protocol.process "rot13-filter.pl clean smudge" &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
test_config_global filter.protocol.required true &&
rm -rf repo &&
mkdir repo &&
(
cd repo &&
git init &&
echo "*.r filter=protocol" >.gitattributes &&
git add . &&
git commit -m "test commit 1" &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
git branch empty-branch &&
cp "$TEST_ROOT/test.o" test.r &&
cp "$TEST_ROOT/test2.o" test2.r &&
mkdir testsubdir &&
cp "$TEST_ROOT/test3 '\''sq'\'',\$x=.o" "testsubdir/test3 '\''sq'\'',\$x=.r" &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
>test4-empty.r &&
S=$(file_size test.r) &&
S2=$(file_size test2.r) &&
S3=$(file_size "testsubdir/test3 '\''sq'\'',\$x=.r") &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
filter_git add . &&
cat >expected.log <<-EOF &&
START
init handshake complete
IN: clean test.r $S [OK] -- OUT: $S . [OK]
IN: clean test2.r $S2 [OK] -- OUT: $S2 . [OK]
IN: clean test4-empty.r 0 [OK] -- OUT: 0 [OK]
IN: clean testsubdir/test3 '\''sq'\'',\$x=.r $S3 [OK] -- OUT: $S3 . [OK]
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
STOP
EOF
test_cmp_count expected.log rot13-filter.log &&
git commit -m "test commit 2" &&
rm -f test2.r "testsubdir/test3 '\''sq'\'',\$x=.r" &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
filter_git checkout --quiet --no-progress . &&
cat >expected.log <<-EOF &&
START
init handshake complete
IN: smudge test2.r $S2 [OK] -- OUT: $S2 . [OK]
IN: smudge testsubdir/test3 '\''sq'\'',\$x=.r $S3 [OK] -- OUT: $S3 . [OK]
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
STOP
EOF
test_cmp_exclude_clean expected.log rot13-filter.log &&
filter_git checkout --quiet --no-progress empty-branch &&
cat >expected.log <<-EOF &&
START
init handshake complete
IN: clean test.r $S [OK] -- OUT: $S . [OK]
STOP
EOF
test_cmp_exclude_clean expected.log rot13-filter.log &&
filter_git checkout --quiet --no-progress master &&
cat >expected.log <<-EOF &&
START
init handshake complete
IN: smudge test.r $S [OK] -- OUT: $S . [OK]
IN: smudge test2.r $S2 [OK] -- OUT: $S2 . [OK]
IN: smudge test4-empty.r 0 [OK] -- OUT: 0 [OK]
IN: smudge testsubdir/test3 '\''sq'\'',\$x=.r $S3 [OK] -- OUT: $S3 . [OK]
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
STOP
EOF
test_cmp_exclude_clean expected.log rot13-filter.log &&
test_cmp_committed_rot13 "$TEST_ROOT/test.o" test.r &&
test_cmp_committed_rot13 "$TEST_ROOT/test2.o" test2.r &&
test_cmp_committed_rot13 "$TEST_ROOT/test3 '\''sq'\'',\$x=.o" "testsubdir/test3 '\''sq'\'',\$x=.r"
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
)
'
test_expect_success PERL 'required process filter takes precedence' '
test_config_global filter.protocol.clean false &&
test_config_global filter.protocol.process "rot13-filter.pl clean" &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
test_config_global filter.protocol.required true &&
rm -rf repo &&
mkdir repo &&
(
cd repo &&
git init &&
echo "*.r filter=protocol" >.gitattributes &&
cp "$TEST_ROOT/test.o" test.r &&
S=$(file_size test.r) &&
# Check that the process filter is invoked here
filter_git add . &&
cat >expected.log <<-EOF &&
START
init handshake complete
IN: clean test.r $S [OK] -- OUT: $S . [OK]
STOP
EOF
test_cmp_count expected.log rot13-filter.log
)
'
test_expect_success PERL 'required process filter should be used only for "clean" operation only' '
test_config_global filter.protocol.process "rot13-filter.pl clean" &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
rm -rf repo &&
mkdir repo &&
(
cd repo &&
git init &&
echo "*.r filter=protocol" >.gitattributes &&
cp "$TEST_ROOT/test.o" test.r &&
S=$(file_size test.r) &&
filter_git add . &&
cat >expected.log <<-EOF &&
START
init handshake complete
IN: clean test.r $S [OK] -- OUT: $S . [OK]
STOP
EOF
test_cmp_count expected.log rot13-filter.log &&
rm test.r &&
filter_git checkout --quiet --no-progress . &&
# If the filter would be used for "smudge", too, we would see
# "IN: smudge test.r 57 [OK] -- OUT: 57 . [OK]" here
cat >expected.log <<-EOF &&
START
init handshake complete
STOP
EOF
test_cmp_exclude_clean expected.log rot13-filter.log
)
'
test_expect_success PERL 'required process filter should process multiple packets' '
test_config_global filter.protocol.process "rot13-filter.pl clean smudge" &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
test_config_global filter.protocol.required true &&
rm -rf repo &&
mkdir repo &&
(
cd repo &&
git init &&
# Generate data requiring 1, 2, 3 packets
S=65516 && # PKTLINE_DATA_MAXLEN -> Maximal size of a packet
generate_random_characters $(($S )) 1pkt_1__.file &&
generate_random_characters $(($S +1)) 2pkt_1+1.file &&
generate_random_characters $(($S*2-1)) 2pkt_2-1.file &&
generate_random_characters $(($S*2 )) 2pkt_2__.file &&
generate_random_characters $(($S*2+1)) 3pkt_2+1.file &&
for FILE in "$TEST_ROOT"/*.file
do
cp "$FILE" . &&
rot13.sh <"$FILE" >"$FILE.rot13"
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
done &&
echo "*.file filter=protocol" >.gitattributes &&
filter_git add *.file .gitattributes &&
cat >expected.log <<-EOF &&
START
init handshake complete
IN: clean 1pkt_1__.file $(($S )) [OK] -- OUT: $(($S )) . [OK]
IN: clean 2pkt_1+1.file $(($S +1)) [OK] -- OUT: $(($S +1)) .. [OK]
IN: clean 2pkt_2-1.file $(($S*2-1)) [OK] -- OUT: $(($S*2-1)) .. [OK]
IN: clean 2pkt_2__.file $(($S*2 )) [OK] -- OUT: $(($S*2 )) .. [OK]
IN: clean 3pkt_2+1.file $(($S*2+1)) [OK] -- OUT: $(($S*2+1)) ... [OK]
STOP
EOF
test_cmp_count expected.log rot13-filter.log &&
rm -f *.file &&
filter_git checkout --quiet --no-progress -- *.file &&
cat >expected.log <<-EOF &&
START
init handshake complete
IN: smudge 1pkt_1__.file $(($S )) [OK] -- OUT: $(($S )) . [OK]
IN: smudge 2pkt_1+1.file $(($S +1)) [OK] -- OUT: $(($S +1)) .. [OK]
IN: smudge 2pkt_2-1.file $(($S*2-1)) [OK] -- OUT: $(($S*2-1)) .. [OK]
IN: smudge 2pkt_2__.file $(($S*2 )) [OK] -- OUT: $(($S*2 )) .. [OK]
IN: smudge 3pkt_2+1.file $(($S*2+1)) [OK] -- OUT: $(($S*2+1)) ... [OK]
STOP
EOF
test_cmp_exclude_clean expected.log rot13-filter.log &&
for FILE in *.file
do
test_cmp_committed_rot13 "$TEST_ROOT/$FILE" $FILE
done
)
'
test_expect_success PERL 'required process filter with clean error should fail' '
test_config_global filter.protocol.process "rot13-filter.pl clean smudge" &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
test_config_global filter.protocol.required true &&
rm -rf repo &&
mkdir repo &&
(
cd repo &&
git init &&
echo "*.r filter=protocol" >.gitattributes &&
cp "$TEST_ROOT/test.o" test.r &&
echo "this is going to fail" >clean-write-fail.r &&
echo "content-test3-subdir" >test3.r &&
test_must_fail git add .
)
'
test_expect_success PERL 'process filter should restart after unexpected write failure' '
test_config_global filter.protocol.process "rot13-filter.pl clean smudge" &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
rm -rf repo &&
mkdir repo &&
(
cd repo &&
git init &&
echo "*.r filter=protocol" >.gitattributes &&
cp "$TEST_ROOT/test.o" test.r &&
cp "$TEST_ROOT/test2.o" test2.r &&
echo "this is going to fail" >smudge-write-fail.o &&
cp smudge-write-fail.o smudge-write-fail.r &&
S=$(file_size test.r) &&
S2=$(file_size test2.r) &&
SF=$(file_size smudge-write-fail.r) &&
git add . &&
rm -f *.r &&
rm -f rot13-filter.log &&
git checkout --quiet --no-progress . 2>git-stderr.log &&
grep "smudge write error at" git-stderr.log &&
grep "error: external filter" git-stderr.log &&
cat >expected.log <<-EOF &&
START
init handshake complete
IN: smudge smudge-write-fail.r $SF [OK] -- OUT: $SF [WRITE FAIL]
START
init handshake complete
IN: smudge test.r $S [OK] -- OUT: $S . [OK]
IN: smudge test2.r $S2 [OK] -- OUT: $S2 . [OK]
STOP
EOF
test_cmp_exclude_clean expected.log rot13-filter.log &&
test_cmp_committed_rot13 "$TEST_ROOT/test.o" test.r &&
test_cmp_committed_rot13 "$TEST_ROOT/test2.o" test2.r &&
# Smudge failed
! test_cmp smudge-write-fail.o smudge-write-fail.r &&
rot13.sh <smudge-write-fail.o >expected &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
git cat-file blob :smudge-write-fail.r >actual &&
test_cmp expected actual
)
'
test_expect_success PERL 'process filter should not be restarted if it signals an error' '
test_config_global filter.protocol.process "rot13-filter.pl clean smudge" &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
rm -rf repo &&
mkdir repo &&
(
cd repo &&
git init &&
echo "*.r filter=protocol" >.gitattributes &&
cp "$TEST_ROOT/test.o" test.r &&
cp "$TEST_ROOT/test2.o" test2.r &&
echo "this will cause an error" >error.o &&
cp error.o error.r &&
S=$(file_size test.r) &&
S2=$(file_size test2.r) &&
SE=$(file_size error.r) &&
git add . &&
rm -f *.r &&
filter_git checkout --quiet --no-progress . &&
cat >expected.log <<-EOF &&
START
init handshake complete
IN: smudge error.r $SE [OK] -- OUT: 0 [ERROR]
IN: smudge test.r $S [OK] -- OUT: $S . [OK]
IN: smudge test2.r $S2 [OK] -- OUT: $S2 . [OK]
STOP
EOF
test_cmp_exclude_clean expected.log rot13-filter.log &&
test_cmp_committed_rot13 "$TEST_ROOT/test.o" test.r &&
test_cmp_committed_rot13 "$TEST_ROOT/test2.o" test2.r &&
test_cmp error.o error.r
)
'
test_expect_success PERL 'process filter abort stops processing of all further files' '
test_config_global filter.protocol.process "rot13-filter.pl clean smudge" &&
convert: add filter.<driver>.process option Git's clean/smudge mechanism invokes an external filter process for every single blob that is affected by a filter. If Git filters a lot of blobs then the startup time of the external filter processes can become a significant part of the overall Git execution time. In a preliminary performance test this developer used a clean/smudge filter written in golang to filter 12,000 files. This process took 364s with the existing filter mechanism and 5s with the new mechanism. See details here: https://github.com/github/git-lfs/pull/1382 This patch adds the `filter.<driver>.process` string option which, if used, keeps the external filter process running and processes all blobs with the packet format (pkt-line) based protocol over standard input and standard output. The full protocol is explained in detail in `Documentation/gitattributes.txt`. A few key decisions: * The long running filter process is referred to as filter protocol version 2 because the existing single shot filter invocation is considered version 1. * Git sends a welcome message and expects a response right after the external filter process has started. This ensures that Git will not hang if a version 1 filter is incorrectly used with the filter.<driver>.process option for version 2 filters. In addition, Git can detect this kind of error and warn the user. * The status of a filter operation (e.g. "success" or "error) is set before the actual response and (if necessary!) re-set after the response. The advantage of this two step status response is that if the filter detects an error early, then the filter can communicate this and Git does not even need to create structures to read the response. * All status responses are pkt-line lists terminated with a flush packet. This allows us to send other status fields with the same protocol in the future. Helped-by: Martin-Louis Bright <mlbright@gmail.com> Reviewed-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Lars Schneider <larsxschneider@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-10-17 07:20:37 +08:00
rm -rf repo &&
mkdir repo &&
(
cd repo &&
git init &&
echo "*.r filter=protocol" >.gitattributes &&
cp "$TEST_ROOT/test.o" test.r &&
cp "$TEST_ROOT/test2.o" test2.r &&
echo "error this blob and all future blobs" >abort.o &&
cp abort.o abort.r &&
SA=$(file_size abort.r) &&
git add . &&
rm -f *.r &&
# Note: This test assumes that Git filters files in alphabetical
# order ("abort.r" before "test.r").
filter_git checkout --quiet --no-progress . &&
cat >expected.log <<-EOF &&
START
init handshake complete
IN: smudge abort.r $SA [OK] -- OUT: 0 [ABORT]
STOP
EOF
test_cmp_exclude_clean expected.log rot13-filter.log &&
test_cmp "$TEST_ROOT/test.o" test.r &&
test_cmp "$TEST_ROOT/test2.o" test2.r &&
test_cmp abort.o abort.r
)
'
test_expect_success PERL 'invalid process filter must fail (and not hang!)' '
test_config_global filter.protocol.process cat &&
test_config_global filter.protocol.required true &&
rm -rf repo &&
mkdir repo &&
(
cd repo &&
git init &&
echo "*.r filter=protocol" >.gitattributes &&
cp "$TEST_ROOT/test.o" test.r &&
test_must_fail git add . 2>git-stderr.log &&
grep "does not support filter protocol version" git-stderr.log
)
'
test_done