Pool scoping issues on callback functions in the Ruby bindings (was: seeking 1.3.0 RC shepherd)

[David James <ja...@gmail.com>]

On 10/13/05, David James <ja...@gmail.com> wrote:
> On 10/13/05, David Anderson <da...@calixo.net> wrote:
> > >  * The Ruby bindings seg fault in their test suite if you build with
> > >  "--enable-pool-debug=yes". John Szakmeister, in this message
> > >  http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgNo=106738
> > >  starts it off, then David James asks Kouhei to take a look here:
> > >  http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgNo=106779
> > >
> > >  [Meta comment: Should it worry us that there's basically only person
> > >  who knows how to keep the SWIG-Ruby bindings in shape? We have a
> > >  minimum two-person-per-language policy when it comes to translations;
> > >  shouldn't we at least have the same for the bindings? Kouhei, this
> > >  is not a complaint about you, of course; you're doing great stuff,
> > >  it's just a question of what would happen if you took a vacation. :-)
> > >  ]
> > It would be nice, to ensure the long-term stability of the Ruby
> > bindings.  FWIW, I'm +1 on stating that a minimum two-devs-per-language
> > in bindings is very strongly desired, if not mandatory.
>
> Are there any budding Ruby or Python developers here?
>
> It would definitely be great to have more bindings developers. If
> anyone is interested in learning more about Subversion, the bindings
> are a great place to start. Feel free to email me and the list if you
> have any questions, and we'll do our best to help.
>
> Still, I don't think we need to worry about the Ruby bindings. The
> Ruby bindings are in excellent shape: they provide extensive coverage
> of the Subversion APIs through an intuitive Ruby interface. They also
> have an extensive test suite, which will forever ensure their
> correctness as we continue to improve Subversion in the future.
>
> I don't think the Ruby issues with the "enable-pool-debug" option are
> a blocker for 1.3.0. If Kouhei and I don't finish the fix on time for
> this release, we can release the fix with 1.3.1.
I've looked into this closer: It looks like the Ruby bindings are
being hit by pool scoping issues, similar to those which affected the
Subversion 1.2 Python bindings on Issue #2172.  I'm working on porting
my Python fix (from r16398) to Ruby.

Cheers,

David


--
David James -- http://www.cs.toronto.edu/~james