[bug#55517] [PATCH]: gnu: emacs-magit: Drop the libgit backend.

[Zhu Zihao <all_but_last@163.com>]

[-- Attachment #1: Type: text/plain, Size: 2283 bytes --]


Maxim Cournoyer <maxim.cournoyer@gmail.com> writes:

> Hi Zhu,
>
> Zhu Zihao <all_but_last@163.com> writes:
>
>> Fix the typo in commit message
>>
>>>From 8a2de5764a67bea1cbf789c8d6ce0ab0878ec40b Mon Sep 17 00:00:00 2001
>> From: Zhu Zihao <all_but_last@163.com>
>> Date: Thu, 19 May 2022 13:01:46 +0800
>> Subject: [PATCH] gnu: emacs-magit: Drop the libgit backend.
>>
>> Drop the libgit backend for following reasons:
>>
>> 1. The libgit backend of Magitis very incomplete. There's almost no benefits,
>> but extra maintenance efforts.
>
> Could you elaborate about "almost no benefits"?  Even if it's a small
> benefit, if it makes it faster, I'd still keep it since Magit is at
> times excruciatingly slow.

It's slow? AFAIK magit user always complain that magit on Windows system
is slow. The process creation on *nix is much cheaper than Windows, so
Magit is not slow on *nix.

I've elaborated the reason in
https://debbugs.gnu.org/cgi/bugreport.cgi?bug=55427.

The libgit backend of Magit is almost nothing, only one function is
implemented. `magit-bare-repo-p` will be used in magit-toplevel which is
called frequently in the magit source code. But this function is cached.

You can still say it's faster, but it's fair to say the libgit backend
doesn't worth the maintenance effort due to the inactive status of
upstream.

>> 2. The libgit backend of Magit can be considered as an extra package. And it's
>> still in Proof-Of-Concept status so its quaility doesn't satisfy the
>> requirement of Guix package.
>
> There's not really any "quality requirements" for Guix package; we
> package stable releases when they're available; the rest is up to us
> Guix users.

If you really wants to keep the libgit backend. I'd suggest to make it a
dedicated package. If emacs-libgit is broken (again), magit is still usable. 

> Do you happen to have a link to some exchange where Magit authors would
> recommend against using the libgit library at this point in time to
> speed up Magit?  Otherwise, I'm not convinced.

No, I've said above. The libgit backend is very very very imcomplete.

>
> Thanks,
>
> Maxim


-- 
Retrieve my PGP public key:

  gpg --recv-keys D47A9C8B2AE3905B563D9135BE42B352A9F6821F

Zihao

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 255 bytes --]