[pytest-dev] removing assertion reinterpretation and making the assertion rewriter a vendored package

[Ronny Pfannschmidt]

the only use-case reinterpret currently has, is fallback for when the
import hook didnt trigger

On Thu, Apr 14, 2016 at 6:24 AM, Florian Bruhin <me at the-compiler.org> wrote:

> * Floris Bruynooghe <flub at devork.be> [2016-04-13 22:58:59 +0100]:
> > Hello,
> >
> > On 13 April 2016 at 13:44, Ronny Pfannschmidt <rpfannsc at redhat.com>
> wrote:
> > > currently we still have a monkey-patch in python that causes
> interesting
> > > issues in c interaction,
> > >
> > > namely the replacement of the Built-in Assertion error by the rewrite
> module
> > >
> > > its highly unlikely we can ever make it act correct, so i propose that
> > > instead we just remove the assertion re-interpreter
> >
> > This sounds good to me. We're currently not maintaining that code
> > properly anyway and it has been 2nd class for a long time. My biggest
> > worry is that occasionally we still say "disable rewrite" as a
> > workaround, but there's still plain so that's probably ok.
> >
> > I assume dropping this is something for 3.0?
> >
> > > while doing that i'd also like to turn the assertion rewriter into a
> package
> > > exposing a minimal api
> > > that way its releases can be done more in lock with python releases
> > >
> > > also its possible for others to use it and contribute to it
> >
> > This I am less keen on. Most code in the re-write package is very
> > py.test specific and often bugs about reporting involve code changes
> > here. We're slowly moving things from py to pytest to avoid this pain
> > so introducing it here seems counter-intuitive.
> >
> > I'm not sure which projects do want to use the assert replacing of
> > py.test. Do you have a use-case in mind?
> >
> > Lastly, in the life of rewrite we've had just one case where the AST
> > changed and we failed to keep up. I don't think this is such a heavy
> > burden and our release process is ever getting better (thanks!) so
> > should be less of an issue next time.
>
> I fully agree with Floris. Dropping reinterpret sounds like a good
> idea as I don't see any usecase for it anymore.
>
> And I also agree that splitting off assertion rewriting is probably
> more trouble than it's worth.
>
> Florian
>
> --
>     http://www.the-compiler.org | me at the-compiler.org (Mail/XMPP)
>        GPG: 916E B0C8 FD55 A072 | http://the-compiler.org/pubkey.asc
>      I'm running a crowdfunding to work on my FOSS-project full-time:
>                   http://igg.me/at/qutebrowser
>
> _______________________________________________
> pytest-dev mailing list
> pytest-dev at python.org
> https://mail.python.org/mailman/listinfo/pytest-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pytest-dev/attachments/20160414/8c7beb55/attachment.html>