Zsh Mailing List Archive
Messages sorted by:
Reverse Date,
Date,
Thread,
Author
Re: [PATCH] Improve vcs_info example for ahead/behind git commits
- X-seq: zsh-workers 48313
- From: Tim Lee <progscriptclone@xxxxxxxxx>
- To: Daniel Shahaf <d.s@xxxxxxxxxxxxxxxxxx>
- Cc: zsh-workers@xxxxxxx
- Subject: Re: [PATCH] Improve vcs_info example for ahead/behind git commits
- Date: Mon, 29 Mar 2021 18:39:27 +0800
- Archived-at: <https://zsh.org/workers/48313>
- Archived-at: <http://www.zsh.org/sympa/arcsearch_id/zsh-workers/2021-03/20210329103927.qzky3fyevptfflvj%40home-guest>
- In-reply-to: <ff17d1f9-d6ff-49bb-83d2-de0978ec61d4@www.fastmail.com>
- List-id: <zsh-workers.zsh.org>
- References: <20210328213628.7rwz62bq5p2isd6b@home-guest> <20210329070611.GM18178@tarpaulin.shahaf.local2> <20210329093056.fegqam75id4yfkiq@home-guest> <ff17d1f9-d6ff-49bb-83d2-de0978ec61d4@www.fastmail.com>
> > > > diff --git a/Misc/vcs_info-examples b/Misc/vcs_info-examples
> > > > index 94b8a7b5e..36d4d3bf8 100644
> > > > --- a/Misc/vcs_info-examples
> > > > +++ b/Misc/vcs_info-examples
> > > > @@ -179,14 +179,18 @@ function +vi-git-st() {
> > > > local ahead behind
> > > > local -a gitstatus
> > > >
> > > > - # for git prior to 1.7
> > > > + # for git prior to 1.7.0
> > > > # ahead=$(git rev-list origin/${hook_com[branch]}..HEAD | wc -l)
> > > > - ahead=$(git rev-list ${hook_com[branch]}@{upstream}..HEAD 2>/dev/null | wc -l)
> > > > + # for git 1.7.0 and 1.7.1
> > > > + # ahead=$(git rev-list @{upstream}..HEAD 2>/dev/null | wc -l)
> > > > + ahead=$(git rev-list --count @{upstream}..HEAD 2>/dev/null)
> > > > (( $ahead )) && gitstatus+=( "+${ahead}" )
> > >
> > > The version of this function in my zshrc starts like this:
> > > .
> > > git rev-parse @{upstream} >/dev/null 2>&1 || return 0
> > > local -a x=( $(git rev-list --left-right --count HEAD...@{upstream} ) )
> >
> > The snippet below displays N/M. How would you make it display +N/-M
> > instead?
>
> Change the last line to hook_com[misc]="+${x[1]} -${x[2]}".
But hook_com[misc]="+${x[1]} -${x[2]}" would display a '+0' and/or '-0'
when the local git branch is not ahead-of or behind the remote branch.
The existing example does not display '+0' or '-0'.
> > git rev-parse @{upstream} >/dev/null 2>&1 || return 0
> > local -a x=( $(git rev-list --left-right --count HEAD...@{upstream} ) )
> > hook_com[misc]="${(j:/:)x}"
> >
> > > The first line is to handle a detached HEAD, I think, and should
> > > presumably be added?
> >
> > Why not do this instead:
> >
> > local -a x=( $(git rev-list --left-right --count HEAD...@{upstream} 2>/dev/null ) )
>
> You mean, why not discard stderr? Out of general principles ("Errors
> shouldn't be silenced"), plus the fact that I've never had that line
> generate an error that needed to be silenced — not with the «git
> rev-parse … || return» right above it.
>
> Or do you mean that line instead of the git-rev-parse(1) call too?
Yes, I mean: instead of this:
git rev-parse @{upstream} >/dev/null 2>&1 || return 0
local -a x=( $(git rev-list --left-right --count HEAD...@{upstream} ) )
Replace both of the lines above with this:
local -a x=( $(git rev-list --left-right --count HEAD...@{upstream} 2>/dev/null ) )
> First, that would run the rest of the function even though $x might be
> an empty array rather than a two-element one. Two, suppose there is an
> error after resolving «@{upstream}» to a commit but before rev-list
> finishes (for instance, an error while walking the commits history).
> With an explicit git-rev-parse(1) call that error would be reported, but
> if the rev-parse and the history walking are lumped into one step any
> failure from which is treated the same way, history walking failures
> would be masked — even though they're probably interesting.
>
> > ---
> > Note: when replying to my email, please ensure that my email address is
> > included in the 'To:' field, since I am not subscribed to the mailing
> > list.
>
> Ack, but for future reference, the customary signature delimiter is "-- "
> (ASCII 0x2D 0x2D 0x20). Many MUAs syntax-highlight that, for example.
Thank you for all this information. I've learned so much today. :-)
Messages sorted by:
Reverse Date,
Date,
Thread,
Author