| Message ID | 20181122105836.GA36193@retiro.local (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Document change in format of raw diff output format | expand |
On Thu, Nov 22, 2018 at 11:58:36AM +0100, Greg Hurrell wrote: > I was troubleshooting some breakage in some code that consumes the > output of `git log --raw` and looking on two machines with different > versions of Git just now I discovered the output format has changed > somewhere between v2.14.5: > > :000000 100644 000000000... 9773b7718... A content/snippets/1157.md > > and v2.19.0: > > :000000 100644 000000000 9773b7718 A content/snippets/1157.md > > A quick search turns up some patches related to the > GIT_PRINT_SHA1_ELLIPSIS env variable, which can be used to force the > old output format, and which landed in v2.16.0, I think. Yes. The actual commit that flipped the default is 7cb6ac1e4b (diff: diff_aligned_abbrev: remove ellipsis after abbreviated SHA-1 value, 2017-12-03). There's more discussion of the possibility of breakage in this subthread: https://public-inbox.org/git/83D263E58ABD46188756D41FE311E469@PhilipOakley/ > Does it sound right that we should update the documentation in > diff-format.txt to show what the new output format is? The examples > all show the old output format, which isn't produced by default any > more. Yes, we should definitely update the documentation to show the modern format. I think that was just an oversight in the original series. > diff --git a/Documentation/diff-format.txt b/Documentation/diff-format.txt > index 706916c94c..33776459d0 100644 > --- a/Documentation/diff-format.txt > +++ b/Documentation/diff-format.txt > @@ -26,12 +26,12 @@ line per changed file. > An output line is formatted this way: > > ------------------------------------------------ > -in-place edit :100644 100644 bcd1234... 0123456... M file0 > -copy-edit :100644 100644 abcd123... 1234567... C68 file1 file2 > -rename-edit :100644 100644 abcd123... 1234567... R86 file1 file3 > -create :000000 100644 0000000... 1234567... A file4 > -delete :100644 000000 1234567... 0000000... D file5 > -unmerged :000000 000000 0000000... 0000000... U file6 > +in-place edit :100644 100644 bcd123456 012345678 M file0 > +copy-edit :100644 100644 abcd12345 123456789 C68 file1 file2 > +rename-edit :100644 100644 abcd12345 123456789 R86 file1 file3 > +create :000000 100644 000000000 123456789 A file4 > +delete :100644 000000 123456789 000000000 D file5 > +unmerged :000000 000000 000000000 000000000 U file6 > ------------------------------------------------ Yeah, this looks like an improvement. I think in general that we'd continue to show 7 characters now, just without the extra dots (though it's auto-scaled based on the number of objects in the repo these days, so it's not even really a constant). > That is, from the left to the right: > @@ -75,7 +75,7 @@ and it is out of sync with the index. > Example: > > ------------------------------------------------ > -:100644 100644 5be4a4...... 000000...... M file.c > +:100644 100644 5be4a4abc 000000000 M file.c > ------------------------------------------------ I'm not even sure what this original was trying to show. I don't think we ever produced that any dots. :) Thanks for noticing. -Peff PS As you noticed, "git log" we don't promise that git-log output will never change between versions. For machine-consumption you probably want to use plumbing like "git rev-list | git diff-tree --stdin", which produces unabbreviated hashes.
Jeff King wrote: > On Thu, Nov 22, 2018 at 11:58:36AM +0100, Greg Hurrell wrote: > > > diff --git a/Documentation/diff-format.txt b/Documentation/diff-format.txt > > index 706916c94c..33776459d0 100644 > > --- a/Documentation/diff-format.txt > > +++ b/Documentation/diff-format.txt > > @@ -26,12 +26,12 @@ line per changed file. > > An output line is formatted this way: > > > > ------------------------------------------------ > > -in-place edit :100644 100644 bcd1234... 0123456... M file0 > > -copy-edit :100644 100644 abcd123... 1234567... C68 file1 file2 > > -rename-edit :100644 100644 abcd123... 1234567... R86 file1 file3 > > -create :000000 100644 0000000... 1234567... A file4 > > -delete :100644 000000 1234567... 0000000... D file5 > > -unmerged :000000 000000 0000000... 0000000... U file6 > > +in-place edit :100644 100644 bcd123456 012345678 M file0 > > +copy-edit :100644 100644 abcd12345 123456789 C68 file1 file2 > > +rename-edit :100644 100644 abcd12345 123456789 R86 file1 file3 > > +create :000000 100644 000000000 123456789 A file4 > > +delete :100644 000000 123456789 000000000 D file5 > > +unmerged :000000 000000 000000000 000000000 U file6 > > ------------------------------------------------ > > Yeah, this looks like an improvement. > > I think in general that we'd continue to show 7 characters now, just > without the extra dots (though it's auto-scaled based on the number of > objects in the repo these days, so it's not even really a constant). That's funny. I looked at the output on (what I thought was) a small repo and it was showing me 9-character abbreviated hashes. I guess I just got lucky. Tested on a basically empty repo and 7 does look to be the default. > PS As you noticed, "git log" we don't promise that git-log output will > never change between versions. For machine-consumption you probably > want to use plumbing like "git rev-list | git diff-tree --stdin", > which produces unabbreviated hashes. Thanks for the tip. My mistake was thinking that the `--raw` made the `git log` output somehow more plumbing-ish, but I've gone ahead and switched to using git-rev-list plus git-diff-tree instead. Anyway, patch follows.
diff --git a/Documentation/diff-format.txt b/Documentation/diff-format.txt index 706916c94c..33776459d0 100644 --- a/Documentation/diff-format.txt +++ b/Documentation/diff-format.txt @@ -26,12 +26,12 @@ line per changed file. An output line is formatted this way: ------------------------------------------------ -in-place edit :100644 100644 bcd1234... 0123456... M file0 -copy-edit :100644 100644 abcd123... 1234567... C68 file1 file2 -rename-edit :100644 100644 abcd123... 1234567... R86 file1 file3 -create :000000 100644 0000000... 1234567... A file4 -delete :100644 000000 1234567... 0000000... D file5 -unmerged :000000 000000 0000000... 0000000... U file6 +in-place edit :100644 100644 bcd123456 012345678 M file0 +copy-edit :100644 100644 abcd12345 123456789 C68 file1 file2 +rename-edit :100644 100644 abcd12345 123456789 R86 file1 file3 +create :000000 100644 000000000 123456789 A file4 +delete :100644 000000 123456789 000000000 D file5 +unmerged :000000 000000 000000000 000000000 U file6 ------------------------------------------------ That is, from the left to the right: @@ -75,7 +75,7 @@ and it is out of sync with the index. Example: ------------------------------------------------ -:100644 100644 5be4a4...... 000000...... M file.c +:100644 100644 5be4a4abc 000000000 M file.c ------------------------------------------------ Without the `-z` option, pathnames with "unusual" characters are @@ -100,7 +100,7 @@ from the format described above in the following way: Example: ------------------------------------------------ -::100644 100644 100644 fabadb8... cc95eb0... 4866510... MM describe.c +::100644 100644 100644 fabadb827 cc95eb0f2 4866510ea MM describe.c ------------------------------------------------ Note that 'combined diff' lists only files which were modified from