[PATCH] Add support for git's mailmap.

Jason A. Smith smithj4 at bnl.gov
Fri Aug 26 22:11:47 CEST 2016


On 08/25/2016 04:12 PM, John Keeping wrote:
> On Thu, Aug 25, 2016 at 03:22:35PM -0400, Jason A. Smith wrote:
>> On 08/24/2016 06:30 PM, John Keeping wrote:
>>> On Wed, Aug 24, 2016 at 05:27:31PM -0400, Jason A. Smith wrote:
>>>> Are you proposing to remove the angle brackets around the email address
>>>> everywhere in cgit strings, to match git, which I think is done in the
>>>> parse_user function? Then use git's map_user directly and create a
>>>> cgit_print_ident function that inserts the angle brackets when printed?
>>>> Then also making local copies of the pointers so freeing the info
>>>> structure doesn't break the mailmap? I think this would be a much larger
>>>> change beyond just supporting the mailmap.
>>>
>>> Yes, pretty much exactly this.  There's only a small number of sites
>>> that actually use author_email and committer_email, most of which are
>>> already printing something, so the fallout shouldn't be too bad.
>>>
>>> The only other cases are opening the email filter, which is why I
>>> suggested wrapping at least part of that process in a new function.
>>
>> So how do you propose splitting the patches to accomplish this? One
>> patch to add mailmap support and another patch to modify how angle
>> brackets are handled in the email addresses? Which patch goes first and
>> who will fix the angle brackets? Are you volunteering? ;)
>
> The cleanup goes first, since it makes the subsequent patch much
> simpler.
>
> I took a quick look at it and will send the patches in reply to this
> message.  It will probably be simplest if you pick them up and send a
> series containing those two patches and the patch for mailmap support.
>

Hi John,

Thanks for your quick help and patients with this, I am new to git 
workflows via email. I re-factored my mailmap patch on top of your two 
patches and will send them in a new thread. I still kept a small 
map_user() wrapper since the callers don't need the string lengths. Let 
me know if this looks good or if there are anymore comments/changes needed.

~Jason


More information about the CGit mailing list