# git difftool: No such file or directory 3 messages from 2021-06-30 to 2021-08-30. Participants: Alan Blotz, Đoàn Trần Công Danh, David Aguilar. Thread: https://gitlist.dev/t/56029 ## Alan Blotz, 2021-06-30 09:38 Subject: git difftool: No such file or directory Message-ID: <176caa21-0232-9f21-7a8d-af136754e354@blotz.org> URL: https://gitlist.dev/e/176caa21-0232-9f21-7a8d-af136754e354%40blotz.org ``` Thank you for filling out a Git bug report! Please answer the following questions to help us understand your issue. What did you do before the bug happened? (Steps to reproduce your issue) mkdir broken-diff cd broken-diff git init mkdir dir1 mkdir dir2 touch dir1/orig cd dir2/ ln -s ../dir1/orig sym cd .. git add dir* git ci -m "init" git checkout -b b git rm dir2/sym git ci -m "remove" git difftool -d master HEAD What did you expect to happen? (Expected behavior) git difftool shall compare both branches. What happened instead? (Actual behavior) git difftool prints an error: fatal: could not open '/tmp/git-difftool.l4UM7e/left/dir2/sym' for writing: No such file or directory Please review the rest of the bug report below. You can delete any lines you don't wish to share. [System Info] git version: git version 2.31.1 cpu: x86_64 no commit associated with this build sizeof-long: 8 sizeof-size_t: 8 shell-path: /bin/sh uname: Linux 5.12.12-300.fc34.x86_64 #1 SMP Fri Jun 18 14:30:51 UTC 2021 x86_64 compiler info: gnuc: 11.0 libc info: glibc: 2.33 $SHELL (typically, interactive shell): /bin/bash [Enabled Hooks] ``` ## Đoàn Trần Công Danh, 2021-07-01 00:35 Subject: Re: git difftool: No such file or directory Message-ID: URL: https://gitlist.dev/e/YN0N6tUeSJM1EBHD%40danh.dev In-Reply-To: <176caa21-0232-9f21-7a8d-af136754e354@blotz.org> ``` On 2021-06-30 11:38:21+0200, Alan Blotz wrote: > Thank you for filling out a Git bug report! > Please answer the following questions to help us understand your issue. > > What did you do before the bug happened? (Steps to reproduce your issue) > > mkdir broken-diff > cd broken-diff > git init > mkdir dir1 > mkdir dir2 > touch dir1/orig > cd dir2/ > ln -s ../dir1/orig sym > cd .. > git add dir* > git ci -m "init" > git checkout -b b > git rm dir2/sym > git ci -m "remove" > git difftool -d master HEAD > > What did you expect to happen? (Expected behavior) > > git difftool shall compare both branches. > > What happened instead? (Actual behavior) > > git difftool prints an error: > > fatal: could not open '/tmp/git-difftool.l4UM7e/left/dir2/sym' for writing: No such file or directory It looks like this behaviour was there from the time difftool was re-written in C in 03831ef7b5, (difftool: implement the functionality in the builtin, 2017-01-19). The perl version didn't have this problem. The perl version create a file in place of that symlink and write the symlink's target into that file. The C version tries to write (and follow?) the symlink. This hack can fix the problem but I'm not sure it's correct: ----8<--- diff --git a/builtin/difftool.c b/builtin/difftool.c index 2115e548a5..737ebb5b1a 100644 --- a/builtin/difftool.c +++ b/builtin/difftool.c @@ -492,12 +492,14 @@ static int run_dir_diff(const char *extcmd, int symlinks, const char *prefix, if (*entry->left) { add_path(&ldir, ldir_len, entry->path); ensure_leading_directories(ldir.buf); - write_file(ldir.buf, "%s", entry->left); + unlink(ldir.buf); + write_file_buf(ldir.buf, entry->left, strlen(entry->left)); } if (*entry->right) { add_path(&rdir, rdir_len, entry->path); ensure_leading_directories(rdir.buf); - write_file(rdir.buf, "%s", entry->right); + unlink(rdir.buf); + write_file_buf(rdir.buf, entry->right, strlen(entry->right)); } } ---->8----- +Cc: Dscho, who wrote the C version. -- Danh ``` ## David Aguilar, 2021-08-30 00:32 Subject: Re: git difftool: No such file or directory Message-ID: URL: https://gitlist.dev/e/CAJDDKr6Z9E6MbQReTXtF3k9V8YJ-mv9swgsazdD0eeAA35UY0A%40mail.gmail.com In-Reply-To: ``` On Wed, Jun 30, 2021 at 5:38 PM Đoàn Trần Công Danh wrote: > > On 2021-06-30 11:38:21+0200, Alan Blotz wrote: > > > Thank you for filling out a Git bug report! > > Please answer the following questions to help us understand your issue. > > > > What did you do before the bug happened? (Steps to reproduce your issue) > > > > mkdir broken-diff > > cd broken-diff > > git init > > mkdir dir1 > > mkdir dir2 > > touch dir1/orig > > cd dir2/ > > ln -s ../dir1/orig sym > > cd .. > > git add dir* > > git ci -m "init" > > git checkout -b b > > git rm dir2/sym > > git ci -m "remove" > > git difftool -d master HEAD > > > > What did you expect to happen? (Expected behavior) > > > > git difftool shall compare both branches. > > > > What happened instead? (Actual behavior) > > > > git difftool prints an error: > > > > fatal: could not open '/tmp/git-difftool.l4UM7e/left/dir2/sym' for writing: No such file or directory > > It looks like this behaviour was there from the time difftool was > re-written in C in 03831ef7b5, (difftool: implement the functionality > in the builtin, 2017-01-19). The perl version didn't have this > problem. > > The perl version create a file in place of that symlink and write the > symlink's target into that file. The C version tries to write (and > follow?) the symlink. > > This hack can fix the problem but I'm not sure it's correct: > ----8<--- > > diff --git a/builtin/difftool.c b/builtin/difftool.c > index 2115e548a5..737ebb5b1a 100644 > --- a/builtin/difftool.c > +++ b/builtin/difftool.c > @@ -492,12 +492,14 @@ static int run_dir_diff(const char *extcmd, int symlinks, const char *prefix, > if (*entry->left) { > add_path(&ldir, ldir_len, entry->path); > ensure_leading_directories(ldir.buf); > - write_file(ldir.buf, "%s", entry->left); > + unlink(ldir.buf); > + write_file_buf(ldir.buf, entry->left, strlen(entry->left)); > } > if (*entry->right) { > add_path(&rdir, rdir_len, entry->path); > ensure_leading_directories(rdir.buf); > - write_file(rdir.buf, "%s", entry->right); > + unlink(rdir.buf); > + write_file_buf(rdir.buf, entry->right, strlen(entry->right)); > } > } > ---->8----- > > +Cc: Dscho, who wrote the C version. Thank you for tracking this down! I re-read the original perl version and this indeed looks like it should do the trick since it's doing the same thing we did in perl. I'm a little ashamed that we didn't have a test case to cover this use case but the reproduction recipe you included looks like a great start. We already have a few tests that use the SYMLINKS test prereq so this would slot in well there. -- David ```