Mercurial 2.0 call for testing

[Matt Mackall]

On Mon, 2011-10-17 at 16:22 -0400, Scott Palmer wrote:
> On 2011-10-17, at 12:33 PM, Matt Mackall wrote:
> 
> > On Mon, 2011-10-17 at 12:01 +0200, Lionel wrote:
> >> Hello All,
> >> 
> >> v2.0RC doesn't work here, still because of issue 2582. Any chance of having
> >> that issue solved for 2.0 (or even in a near future)? We're stuck to v1.8.4
> >> because of it.
> > 
> > I don't really understand what's affected. Does this break Steve's
> > TortoiseHG builds?
> 
> It makes the binaries for Windows unusable via mod_wsgi on Apache for
> one.

I see. You're aware that it's perfectly fine to use any clients 1.0 -
2.0 with any server 1.0 - 2.0, right? Unless you need a very recent
feature like the faster discovery protocol, you probably won't notice
anything change from upgrading your server.

In fact, I would actually recommend against upgrading clients and
servers at the same time. If you do and something breaks, you won't know
which side broke.

>   The C runtime libraries won't load in that scenario due to issues
> with how the Mercurial binaries are built.  I wish I was expert enough
> in the matter to assist, but all I know is that it likely means that
> the appropriate "manifest" is not embedded in the native bits of the
> Windows binaries for Mercurial.

Compiling your own Mercurial for your server should not be prohibitively
hard, provided you've got an appropriate compiler somewhere.

-- 
Mathematics is the supreme nostalgia of our time.