Re: [PATCH v4 06/12] ref-filter: introduce color_atom_parser()

From: Eric Sunshine <sunshine@sunshineco.com>
Date: Feb 7, 2016, 07:53 UTC
Message-ID: <CAPig+cRBTA4TCwW60UubjLWzzSdO3WiGG+bNy6Wh2Y5-=hYjJA@mail.gmail.com>
In-Reply-To: <CAP8UFD0BMNYiL-N=eTwXvyMX4_exwURzufesW_xYktZoVPaoGA@mail.gmail.com>

On Sat, Feb 6, 2016 at 10:51 AM, Christian Couder <christian.couder@gmail.com> wrote:

Show 16 quoted lines

> On Sat, Feb 6, 2016 at 4:20 PM, Karthik Nayak <karthik.188@gmail.com> wrote:
>> Also the error is reported already in color_parse(...), so seems duplicated.
>>
>> git for-each-ref  --format="%(color:sfadf)%(align:middle,30)%(refname)%(end)"
>> error: invalid color value: sfadf
>> fatal: invalid color value: sfadf
>>
>> What would be an ideal way around this?
>
> Maybe it has already been discussed a lot and I missed the discussion,
> but if possible the argument, the parameter or the atom itself might
> just be ignored with a warning instead of dying when an atom argument,
> format or parameter is not recognized, because in the next Git
> versions we might want to add new arguments, formats and parameter and
> it would be sad if old versions of Git die when those new things are
> passed to them.

The current behavior of die()ing is inherited from existing code which Karthik refactored to create ref-filter.c, so it is not a new issue, and old versions of Git are already afflicted. Whether die()ing is desirable or not is unrelated to the current series which is primarily aimed at optimizing and slightly generalizing ref-filter.c.