SWIG versions - summary of previous thread, and moving forward

[Max Bowsher <ma...@ukf.net>]

The overall status coming out of the previous thread, IIUC, is:

 * There's no particular known reason not to increase the maximum
   acceptable version to SWIG 1.3.31 right now.

     - Accordingly, I'll make that change.

     - And furthermore, I suggest we update the release manager
       designated version of SWIG from 1.3.25 to 1.3.31 for Subversion
       1.5.x.

 * We cannot immediately raise our minimum requirement, because the
   Windows builds are known to have problems with anything other than
   1.3.24. NB: the Windows distribtion does _not_ ship pregenerated SWIG
   files in the zipfile, unlike the Unix distribution.

     - Hopefully at some point someone will resolve whatever problem is
       keeping Windows builds from using SWIG >> 1.3.24, and at that
       point we can raise our minimum SWIG version to 1.3.29 initially.


Max.

Re: SWIG versions - summary of previous thread, and moving forward

[Joe Swatosh <jo...@gmail.com>]

Thanks for taking this on Max,

On 9/30/07, Max Bowsher <ma...@ukf.net> wrote:
> The overall status coming out of the previous thread, IIUC, is:
>
>  * There's no particular known reason not to increase the maximum
>   acceptable version to SWIG 1.3.31 right now.
>
>     - Accordingly, I'll make that change.
>
>     - And furthermore, I suggest we update the release manager
>       designated version of SWIG from 1.3.25 to 1.3.31 for Subversion
>       1.5.x.
>
>  * We cannot immediately raise our minimum requirement, because the
>   Windows builds are known to have problems with anything other than
>   1.3.24. NB: the Windows distribtion does _not_ ship pregenerated SWIG
>   files in the zipfile, unlike the Unix distribution.
>
>     - Hopefully at some point someone will resolve whatever problem is
>       keeping Windows builds from using SWIG >> 1.3.24, and at that
>       point we can raise our minimum SWIG version to 1.3.29 initially.
>
AFAIK, it is currently only the Perl bindings that show this issue,
and only since apr_uint64_t was used in an interface that SWIG is
generating a wrapper for.  IOW, I *think* it has been solved for "long
long" just not "unsigned long long."

I hope that clarification is useful.
--
Joe

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org