Re: Specifying package patches in a more convenient form

[ludo@gnu.org (Ludovic Courtès)]

Alex Kost <alezost@gmail.com> skribis:

> Andy Wingo (2016-04-07 13:08 +0300) wrote:
>
>> On Thu 07 Apr 2016 11:52, Alex Kost <alezost@gmail.com> writes:
>>
>>> Eric Bavier (2016-04-06 17:57 +0300) wrote:
>>>
>>>> On Wed, 06 Apr 2016 15:13:47 +0300
>>>> Alex Kost <alezost@gmail.com> wrote:
>>> [...]
>>>>> > +         "1lgghck46p33z3hg8dnl76jryig4fh6d8rhzms837zp7x4hyfkv4"))
>>>>> > +       (patches (map search-patch '("ttfautohint-source-date-epoch.patch")))))
>>>>>
>>>>> Since it's just a single patch, I don't see a reason to use 'map' here.
>>>>
>>>> Just that it's less to change if more patches are added later.  The
>>>> same has been used in other packages.
>>>
>>> I strongly disagree with this policy.  More patches may never be added,
>>> but mapping through a list of a single element looks redundant for me.
>>
>> What if the "patches" field just applied `search-path' to each of the
>> items in the list if the path is not absolute?  Use
>> `absolute-file-name?' to check if this is needed or not.
>
> I think it is a good choice that 'patches' field takes a list of file
> names.  For example, currently a user can do:
>
>   (patches (find-my-patches "package-name"))
>
> With what you suggest, it would not be possible.

It would still be possible, provided ‘find-my-patches’ returns absolute
file names.  But yeah, there would always be this extra pass of
guesswork under the hood.

Ludo’.