Py3kport: Foolscap porting

[Anand B Pillai]

On Thursday 25 July 2019 05:30 PM, tahoe-dev-request at tahoe-lafs.org wrote:
>
>
>
> Building this extra CI system is extra work and that's why I wouldn't pick
> this plan.  Instead, I'd pick a plan that leverages the existing CI
> system.  This is something like the "make one focused, well-defined set of
> changes that add Python 3 support and don't break anything" plan that has
> been proposed and agreed upon for the Tahoe-LAFS codebase itself.  Since
> you're starting from "everything works" and maintaining "everything works"
> the normal CI system continues to provide the desired check.  If you start
> from "everything's broken" and then "fix one thing" you've broken the
> assumptions of normal CI systems and need to fix the mismatch somehow.
Yes, makes sense definitely. I was thinking in terms of making each
tests pass
one by one in Python3 and backporting related changes to Python2. This will
however break other tests and we will be always in a state of flux during
the port. Not a good idea.

I have worked on my changes and adopted a set of initial changes
for a few core modules with the CI tests passing on 2. This is ready for
an early review at

https://github.com/warner/foolscap/pull/52

The PR is from a new py3k_port branch to warner:py3kport_integration branch.
Please do take a look.

>
> Thanks,
> Jean-Paul
>
>
>

-- 
Kind Regards,

--Anand

-----------------------------------
Founder & Director,

Anvetsu Technologies Pvt Ltd (OPC),

https://www.anvetsu.com
https://www.anvetsu.training
-----------------------------------