Re: [PATCH 5/5] drop vcs-svn experiment

[Jeff King <peff@peff.net>]

On Thu, Aug 13, 2020 at 11:11:36AM -0400, Taylor Blau wrote:

> > So it seems like the result is not really usable by anybody in practice.
> > It would be wonderful if somebody wanted to pick up the topic again, and
> > potentially it's worth carrying around for that reason. But the flip
> > side is that people doing tree-wide operations have to deal with this
> > code.  And you can see the list with (replace "HEAD" with this commit as
> > appropriate):
> 
> I have nothing to add to this email other than to say that I think it
> makes sense to remove this. I agree it would be nice if somebody picked
> it up, but given that nothing has really changed since ~8 years ago, I
> think that it's unlikely to happen in the next ~8 years ;).
> 
> At least, it seems unlikely enough that dropping 2MB from the
> installation size is worth it. If someone *is* interested in picking it
> up, I'm sure that they won't have trouble finding this patch and
> reverting it.

TBH I care less about the installation size (which we could easily fix
by just not installing git-remote-testsvn) and more that it's 3000 lines
of basically-dead code. That keeps getting touched for tree-wide
refactors (most recently for strvec, but my -Wunused-parameters series
was the straw that caused me to just delete it instead).

I don't think resurrecting it will be quite as easy as reverting this.
Because the rest of the code will move on, so it will probably need
some fixups, depending on how long it has been. But those fixups are the
exact same changes we're saving regular git developers from doing. So
it's really just gambling that we won't need to do them (and also
shifting the work from everybody onto whoever chooses to work on this).

It would be nice if it could be spun out independently, but it does use
a bunch of data structures from libgit.a, which complicates that.

-Peff