Re: [PATCH/RFC] git-remote-mediawiki: new tool to preview local changes without pushing

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Sun, Jun 09, 2013 at 02:35:45PM +0200, Benoît Person wrote:

> On 9 June 2013 08:08, Jeff King <peff@xxxxxxxx> wrote:
> > I also wonder if it would be useful to be able to specify not only files
> > in the filesystem, but also arbitrary blobs. So in 4b above, you could
> > "git mw preview origin:page.mw" to see the rendered version of what
> > upstream has done.
> 
> Hum, so `git mw preview origin:page.mw` would just do the get
> request to the remote mediawiki, save it locally and - maybe - load it
> in the browser ? Is it really better than just opening the browser and
> typing the right URL ?

Not really, but when you start doing "origin^:page.mw", it may make more
of a difference.

> > If you just care about the remote name and not the name of the local
> > branch, you can just ask for
> >
> >   my $curr_branch = `git symbolic-ref HEAD`;
> >   my $remote = `git config "branch.$curr_branch.remote"`;
> >   my $url = `git config "remote.$remote.url"`;
> >
> > Of course you would want some error checks and probably some chomp()s in
> > there, too.
> 
> The fact is, `git symbolic-ref HEAD` result would have to be parsed in order
> to extract the current branch name like I currently extract the remote name.
> So, is it really better than `git rev-parse --symbolic-full-name @{upstream}` ?

It is, because it is not strictly true that an upstream of
"refs/remotes/foo/*" is for the remote "foo" (though in 99% of cases, it
is). To find the upstream, git looks at branch.$curr_branch.remote, then
calculates the upstream based on the fetch refspecs. Then you try to
undo that by converting it back from the right-hand side of the fetch
refspec into a remote name. It would be much easier to just look at the
config. :)

And yes, you do need the short name of the branch from HEAD, not the
full refname. You can use "git symbolic-ref --short" for that. You also
should check that it returns something useful (i.e., that we are not on
a detached HEAD).

> > That is probably OK as long as there is only one such remote, and it
> > would help the case where you have branched off of a local branch (so
> > your upstream remote is ".").  If there are two mediawiki remotes,
> > though, it would make sense to simply fail, as you don't know which to
> > use. But I'd expect the common case by far to be that you simply have
> > one such remote.
> 
> Well, I thought that `git mw preview` could provide an interactive mode
> where, if the first search fails, it would find all the mediawiki remotes, and
> offers to the user a way to choose the remote ?

That's fine; just as long as we do not silently produce output from an
unknown source when there is ambiguity.

You can do an interactive selection, or even just say something like:

  There are multiple mediawiki remotes, and I do not know which one you
  want to use for your preview. Which of:

    remote1
    remote2

  did you mean? Try using the "-r" option to specify the remote.

and then let the user repeat their command with the "-r" option using
shell history. That saves you from having to write an interactive
component, and teaches the user how to script it.

-Peff
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html




[Index of Archives]     [Linux Kernel Development]     [Gcc Help]     [IETF Annouce]     [DCCP]     [Netdev]     [Networking]     [Security]     [V4L]     [Bugtraq]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux SCSI]     [Fedora Users]