Stop rejecting Summary: Signed-off-by: ... OR patch Phabricator

Herbert Valerio Riedel hvriedel at gmail.com
Fri May 13 07:00:38 UTC 2016


On 2016-05-13 at 00:40:03 +0200, Edward Z. Yang wrote:
> Currently the commit hooks reject commit messages that look like:
>
>     Summary: Signed-off-by: Edward Z. Yang <ezyang at cs.stanford.edu>
>
> Unfortunately, if I do a one line commit message with -s, Phabricator
> will automatically reformat my message to have this.

...and that's exactly why I invested time to write that hook in the
first place, because Phabricator blatantly violates Git convention
regarding the git commit message format. Stock Phabricator insists on
its own "structured" Git commit format, thereby breaking almost every
Git-tool which expects the usual git-trailer grammar.

I've been meaning to fix this (and another related issue of Phabricator
using whitespaces in trailer-keys[1] which unsurpisingly breaks git
tooling as well but for a different reason, as well as the annoying
issue of Signed-off-by lines sometimes confusing Phabricator...*sigh*)
in Phabricator for some time but never got to it, but...

>     2) You should patch Phabricator to stop munging the messages
>     this way.  The message formatting is done server-side so as
>     a client I have no way of changing it, you need to fix it.
>     This URL has some guidance on how to do it:
>         https://secure.phabricator.com/T6055
>
> I'd offer to edit the hook myself but there does not seem to be
> any canonical location where the hooks are versioned.

...I've tried to patch phabricator as suggested in T6055 (and commented
there), but it seems I need to tweak Phabricator's commit msg parser as
well...



 [1]: E.g. "Reviewed By:" instead of the better supported "Reviewed-by:",
      see e.g. https://git.wiki.kernel.org/index.php/CommitMessageConventions
  
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 818 bytes
Desc: not available
URL: <http://mail.haskell.org/pipermail/ghc-devs/attachments/20160513/db6f9fa8/attachment.sig>


More information about the ghc-devs mailing list