Re: Python 3.2 cygport script

[Jason Tishler <jason at tishler dot net>]

Yaakov,

I really appreciate your feedback.  Can you post your WIP Python 3.2
cygport script and patches?  See below for my comments.

On Fri, Jun 15, 2012 at 02:59:42AM -0500, Yaakov (Cygwin/X) wrote:
> On 2012-06-14 13:56, Jason Tishler wrote:
> >See attached for a diff between your python3-3.1.5rc1-1 cygport script
> >and my WIP 3.2 one.  I'm still deciding on which patches to include
> >and/or getting them to apply cleanly, so ignore those diffs for now.

I'm ready to work on the patches now, so any feedback will be very
helpful.

> >I'm most interested in your feedback on my changes to src_install(), so
> >please focus your attention there.  In particular, I'm trying to figure
> >out how to best handle the new ABIFLAGS stuff that was added in 3.2.
> >Any comments will be greatly appreciated.
> 
> Comments inline.
> 
> >-doc_v=3.1.3
> >+doc_v=${PV}
> 
> That's fine for now, but by later point releases the docs tend not
> to be updated.

I'll hardcode the version when necessary.

> Perhaps this should be a separate source package anyway, but that's up
> to you.

I'd rather focus on the other things in my queue right now.

> >+	3.2.3-extension.patch
> 
> I never had this problem because I set obcaseinsensitive=0.  I'm not
> sure what side effects your patch may have.

AFAICT, my patch will only affect Cygwin Python during the build (unless
someone is silly enough to create a directory call python in /usr/bin).
Otherwise, it will never be triggered.

> >+	3.2.3-libpython-abi.patch
> 
> Yep, I had a similar patch locally from an earlier attempt at 3.2.

Thanks for the confirmation that I was on the right track.

> >+	abi=$(${B}/python.exe -c 'import sys; print(sys.abiflags)')
> 
> I recommend you hard-code this (abi=m) right after the slot define.
> Otherwise, if something happens and the ABI changes suddenly, you
> might miss it and inadvertently ship an update which will break
> users' installations.

OK.  FWIW, that is exactly what I had for my first version.

> >+	dosym idle${slot} /usr/bin/idle3
> 
> I think the same can be done with pydoc.

Will do.

> Otherwise this looks good.

Once again, I really appreciate your feedback.

Thanks,
Jason