# [PATCH 2/2] Setup working tree in describe 4 messages from 2019-01-26 to 2019-01-26. Participants: Sebastian Staudt, Duy Nguyen, Jeff King. Thread: https://gitlist.dev/t/50326 ## Sebastian Staudt, 2019-01-26 10:40 Subject: [PATCH 2/2] Setup working tree in describe Message-ID: URL: https://gitlist.dev/e/CA%2BxP2Sax%2BthitfKv4hTtKTYPhfVXJxD_qoJxgkCyZFTzskP-Tw%40mail.gmail.com ``` This ensures the given working tree is used for --dirty and --broken. Signed-off-by: Sebastian Staudt --- builtin/describe.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/builtin/describe.c b/builtin/describe.c index cc118448ee..ba1a0b199b 100644 --- a/builtin/describe.c +++ b/builtin/describe.c @@ -601,6 +601,8 @@ int cmd_describe(int argc, const char **argv, const char *prefix) if (!hashmap_get_size(&names) && !always) die(_("No names found, cannot describe anything.")); + setup_work_tree(); + if (argc == 0) { if (broken) { struct child_process cp = CHILD_PROCESS_INIT; -- 2.20.1 ``` ## Duy Nguyen, 2019-01-26 11:01 Subject: Re: [PATCH 2/2] Setup working tree in describe Message-ID: URL: https://gitlist.dev/e/CACsJy8Df6XM55ExRfCjpUQsv2Vm0cOwrSAyNf6net__uztMOGQ%40mail.gmail.com In-Reply-To: ``` On Sat, Jan 26, 2019 at 5:44 PM Sebastian Staudt wrote: > > This ensures the given working tree is used for --dirty and --broken. > > Signed-off-by: Sebastian Staudt > --- > builtin/describe.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/builtin/describe.c b/builtin/describe.c > index cc118448ee..ba1a0b199b 100644 > --- a/builtin/describe.c > +++ b/builtin/describe.c > @@ -601,6 +601,8 @@ int cmd_describe(int argc, const char **argv, > const char *prefix) > if (!hashmap_get_size(&names) && !always) > die(_("No names found, cannot describe anything.")); > > + setup_work_tree(); This forces worktree's presence in all cases and will die() if worktree is not available. You need to check if broken or dirty is set and only call this function in that case. Though in my opinion it's better to call before we need it in the "if (broke)" and "else if (dirty)" code blocks. That way you don't even need to check if it's "dirty" or "broken". Does "broken" really need this though? If it runs "git diff-index" separately, that command should handle this setup_work_tree() already, or we may need to fix it there, not here. > + > if (argc == 0) { > if (broken) { > struct child_process cp = CHILD_PROCESS_INIT; > -- > 2.20.1 -- Duy ``` ## Sebastian Staudt, 2019-01-26 14:07 Subject: Re: [PATCH 2/2] Setup working tree in describe Message-ID: URL: https://gitlist.dev/e/CA%2BxP2SaJYD%2BOKC18y0zz2V6A9J8rAB0-cxMnrDmWt%3D%3D8v2_uBA%40mail.gmail.com In-Reply-To: ``` Am Sa., 26. Jan. 2019 um 12:01 Uhr schrieb Duy Nguyen : > > On Sat, Jan 26, 2019 at 5:44 PM Sebastian Staudt wrote: > > > > This ensures the given working tree is used for --dirty and --broken. > > > > Signed-off-by: Sebastian Staudt > > --- > > builtin/describe.c | 2 ++ > > 1 file changed, 2 insertions(+) > > > > diff --git a/builtin/describe.c b/builtin/describe.c > > index cc118448ee..ba1a0b199b 100644 > > --- a/builtin/describe.c > > +++ b/builtin/describe.c > > @@ -601,6 +601,8 @@ int cmd_describe(int argc, const char **argv, > > const char *prefix) > > if (!hashmap_get_size(&names) && !always) > > die(_("No names found, cannot describe anything.")); > > > > + setup_work_tree(); > > This forces worktree's presence in all cases and will die() if > worktree is not available. You need to check if broken or dirty is set > and only call this function in that case. > > Though in my opinion it's better to call before we need it in the "if > (broke)" and "else if (dirty)" code blocks. That way you don't even > need to check if it's "dirty" or "broken". Does "broken" really need > this though? If it runs "git diff-index" separately, that command > should handle this setup_work_tree() already, or we may need to fix it > there, not here. > Thanks for your feedback. Are you sure that it will fail without a working tree? Is it even possible to have *no* working tree? I already tested this with some real life examples, e.g. git --git-dir /some/path/.git describe From inside and outside of other repositories. I didn‘t hit any errors so far. > > + > > if (argc == 0) { > > if (broken) { > > struct child_process cp = CHILD_PROCESS_INIT; > > -- > > 2.20.1 > > > > -- > Duy Best regards, Sebastian Am Sa., 26. Jan. 2019 um 12:01 Uhr schrieb Duy Nguyen : > > On Sat, Jan 26, 2019 at 5:44 PM Sebastian Staudt wrote: > > > > This ensures the given working tree is used for --dirty and --broken. > > > > Signed-off-by: Sebastian Staudt > > --- > > builtin/describe.c | 2 ++ > > 1 file changed, 2 insertions(+) > > > > diff --git a/builtin/describe.c b/builtin/describe.c > > index cc118448ee..ba1a0b199b 100644 > > --- a/builtin/describe.c > > +++ b/builtin/describe.c > > @@ -601,6 +601,8 @@ int cmd_describe(int argc, const char **argv, > > const char *prefix) > > if (!hashmap_get_size(&names) && !always) > > die(_("No names found, cannot describe anything.")); > > > > + setup_work_tree(); > > This forces worktree's presence in all cases and will die() if > worktree is not available. You need to check if broken or dirty is set > and only call this function in that case. > > Though in my opinion it's better to call before we need it in the "if > (broke)" and "else if (dirty)" code blocks. That way you don't even > need to check if it's "dirty" or "broken". Does "broken" really need > this though? If it runs "git diff-index" separately, that command > should handle this setup_work_tree() already, or we may need to fix it > there, not here. > > > + > > if (argc == 0) { > > if (broken) { > > struct child_process cp = CHILD_PROCESS_INIT; > > -- > > 2.20.1 > > > > -- > Duy ``` ## Jeff King, 2019-01-26 14:28 Subject: Re: [PATCH 2/2] Setup working tree in describe Message-ID: <20190126142854.GA20692@sigill.intra.peff.net> URL: https://gitlist.dev/e/20190126142854.GA20692%40sigill.intra.peff.net In-Reply-To: ``` On Sat, Jan 26, 2019 at 03:07:54PM +0100, Sebastian Staudt wrote: > Are you sure that it will fail without a working tree? > Is it even possible to have *no* working tree? > > I already tested this with some real life examples, e.g. > > git --git-dir /some/path/.git describe > > From inside and outside of other repositories. > I didn‘t hit any errors so far. Try: git clone --bare . bare.git cd bare.git git describe That works with current versions of Git, but yields "fatal: this operation must be run in a work tree" with your patch. -Peff PS Your patches seem whitespace-damaged. You might want to look into using git-send-email. ```