Re: [patch] add copying of bdb and sqlite dlls to win-tests.py

Hi David,

On 4/2/07, David James <james@cs.toronto.edu> wrote:
> On 4/2/07, Joe Swatosh <joe.swatosh@gmail.com> wrote:
> > Hi DJ
> >
> > On 4/1/07, D.J. Heap <djheap@gmail.com> wrote:
> > > On 4/1/07, Joe Swatosh <joe.swatosh@gmail.com> wrote:
> > > [snip]
> > > >
> > > > Sorry my formatting is off for the patch notes. But, I'm not sure
> > > > this is the best way to handle these issues anyway. But this works
> > > > with VC6 and SWIG 1.3.24 and 1.3.31.
> > > >
> > > > [[[
> > > > trying to get versions of SWIG later than 1.3.24 working on windows.
> > > >
> > > > * build.conf
> > > > added ruby and perl fragments files to the SWIG "check-out" list.
> > > > * build/generator/swig/checkout_swig_header.py
> > > > dink the fragments files to so "long long" won't be used by SWIG
> > > > * subversion/bindings/swig/include/svn_global.swg
> > > > added typemaps for ruby and perl
> > > > ]]]
> > >
> > >
> > > Thanks -- I'll let the swig experts look this over and hopefully get
> > > it applied soon.
> > >
> >
> > >
> >
> > I was scrambling the other day. Should I resend this patch under a
> > better subject line (in hopes of getting the SWIG experts to comment)?
>
> Yes, could you post the new version of your patch with a new subject
> line? This thread is getting a bit long.
>
> I have some comments on your changes to the build generator.
>
> 1) Could you mention the names of the functions you changed in the log
> message? (e.g. checkout, _skip_checkout). If you could also add more
> detail to your log message that would be great. Your log message
> should, for example, mention that the purpose of the change is to
> prevent "long long" from appearing in the generated .c files, which
> breaks VC6.
>
> 2) In the checkout function: It looks like you just copied 'n pasted
> my Python workaround and changed a variable name. The new code you
> wrote is otherwise identical, so can you just update the "if
> statement" and variable name instead of creating a new block?
>
> 3) In _skip_checkout: Why are we always skipping the new
> ruby/rubyfragments.swg and perl5/perlfragments.swg files? Normally, we
> only skip the checkout of files when they are not present in a
> particular version of SWIG (e.g. ruby/rubytracking.swg is not present
> in older versions of swig, and common.swg is not present in newer
> versions of swig).
>
> Regarding your changes to svn_global.swg:
>
> 1) These changes look pretty good, although I don't know a lot about
> Perl internals. If you feel confident about the changes, it might be
> ready to commit, but it would be nice to get some comments from a Perl
> expert.
>
> 2) The indentation in your changes to svn_global.swg looks a bit off
> (some lines are indented 2 spaces, while others are indented 4 spaces
> -- could you fix that?)
>
> 3) In your log message you should mention that the new typemaps are
> for "long long", and are intended to replace the broken SWIG typemaps
> which you deleted by playing with the rubyfragments.swg and
> perlfragments.swg files in the build generator.
>
> Cheers,
>
> David
>

Thanks for your feedback. I should be able to look at this again
before the weekend.

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