On Wed, 2017-07-05 at 12:50 -0700, Junio C Hamano wrote:
> Three things that caught my eyes:
> 
>  - Between "git commit --cleanup=strip" and "git commit --
> cleanup=verbatim",
>    lines that make up this initial instruction section are different.
> 
>  - "git grep 'Please enter the '" finds that this string is subject
>    to translation, so the pattern may not match (in which case it
>    will be a no-op without doing any harm, which is OK).
> 
>  - core.commentChar can be set to something other than '#', so the
>    pattern may not match (I do not offhand know if that may cause a
>    wrong line to match, causing harm, or not).
> 
> As merely an example, it probably is OK to say "this won't work if
> you are not using the C locale, and/or you are using custom
> core.commentChar".  So if we disregard the latter two, I would think
> 
>     sed -e '/^# Please enter the commit message /,/^#$/d'
> 
> may be simpler to reason about to achieve the same goal.  
> 
Thanks for enlightening me about this. I thought sed was greedy with
address spaces the same way it's greedy with regex.

    sed -e '/^# Please enter the commit message /,/^#$/d'


This command does seem to work regardless of the cleanup mode used.

That said, in case my interpretation that "'prepare-commit-msg' hook is
not to be shipped due to it's uselessness" is correct, the reply of
this mail as a whole seems to contradict it.

Should I work on this patch and another related one (he one that
modifies the signature part of the hook) or
should I drop it ?

IOW, would this patch likely make the hook useful again?

-- 
Kaartic

Reply via email to