[PATCH] Fix: prevent gobject._gobject demandimport crash (issue2502)

[Matt Mackall]

On Wed, 2010-11-17 at 20:28 +0100, Alan Franzoni wrote:
> On Wed, Nov 17, 2010 at 8:19 PM, Matt Mackall <mpm at selenic.com> wrote:
> > You should instead extend the blacklist from within your application.
> > Not only will this mean not having to send us patches, but you can be
> > compatible with versions of Mercurial that don't include your patch.
> 
> Possible approaches are being discussed in the ticket:
> 
> http://mercurial.selenic.com/bts/issue2502
> 
> The problem is:
> 
> * the error really arises in a python library (keyring) which is used
> by a mercurial extension (mercurial_keyring)
> * keyring maintainers don't seem prone to accept such a patch since
> they think it's got nothing to do with their lib

And they're obviously right.

> * mercurial_keyring maintainers chose keyring because it's really an
> abstraction library - it lets client code to "forget" what particular
> keyring implementation is using, so it really seems an overkill.

And these guys are obviously wrong. For the same reasons it's not
keyring's responsibility to deal with weird bugs caused by random
applications, it's not Mercurial's job to deal with bugs caused by
combining it with random libraries. The responsibility falls on the
thing that introduces the two: the mercurial_keyring extension.

And it's doubly their responsibility because only fixing it in the
extension can make it be compatible with pre-existing versions of
Mercurial and keyring.

-- 
Mathematics is the supreme nostalgia of our time.