[Piglit] [ANNOUNCE] python 3 is here... and python 2 is staying

[Ilia Mirkin]

On Tue, Feb 9, 2016 at 2:14 PM, Dylan Baker <baker.dylan.c at gmail.com> wrote:
> Quoting Matt Turner (2016-02-09 10:03:59)
>> On Tue, Feb 9, 2016 at 10:02 AM, Ilia Mirkin <imirkin at alum.mit.edu> wrote:
>> > On Tue, Feb 9, 2016 at 12:55 PM, Dylan Baker <baker.dylan.c at gmail.com> wrote:
>> >> Quoting Matt Turner (2016-02-08 21:06:58)
>> >>> On Mon, Feb 8, 2016 at 3:12 PM, Dylan Baker <baker.dylan.c at gmail.com> wrote:
>> >>> > I just pushed patches to hybridize the framework to use either python
>> >>> > 3.3+ or 2.7.
>> >>> >
>> >>> > Everything should keep working as normal. Unless you have python 3.x,
>> >>> > then you'll need to rerun CMake to pick up the changes.
>> >>>
>> >>> There's still a handful of *.py files with #!/usr/bin/env python2:
>> >>>
>> >>> generated_tests/random_ubo-arb_uniform_buffer_object.py:#!/usr/bin/env python2
>> >>> generated_tests/random_ubo.py:#!/usr/bin/env python2
>> >>> generated_tests/random_ubo_trim.py:#!/usr/bin/env python2
>> >>> piglit-print-commands.py:#!/usr/bin/env python2
>> >>> piglit-resume.py:#!/usr/bin/env python2
>> >>> piglit-run.py:#!/usr/bin/env python2
>> >>> piglit-summary-html.py:#!/usr/bin/env python2
>> >>> piglit-summary.py:#!/usr/bin/env python2
>> >>> self-tests/test-installed-piglit-script-imports-correct-framework-module:#!/usr/bin/env
>> >>> python2
>> >>
>> >>
>> >> I thought I'd mentioned that in the commit, but apparently I did only in
>> >> the cover letter.
>> >>
>> >> The random_ubo_stuff is not plugged into anything at the moment, and I
>> >> have a WIP branch for those, but getting the same output from python 2
>> >> and python 3 for generators without numpy is tricky.
>> >>
>> >> The piglit-* command were left intentionally. They're legacy interfaces
>> >> and we don't add new ones when adding new functions to the 'piglit'
>> >> command (piglit summary aggregate, for example). Personally, I'd rather
>> >> get rid of them (other than print-commands, which I should probably
>> >> update).
>> >
>> > Really? I think it's a lot more straightforward to use them than the
>> > "piglit" megarunner thing... not sure why that was added tbh. But
>> > since the regular commands stayed, I didn't object at the time. Should
>> > I have?
>>
>> Yeah, I've continued using the piglit-run.py/piglit-summary.py
>> scripts. I don't know what I would gain by switching, but I would lose
>> the ability to tab complete their names. :)
>
> One of these days I should really finish writing those zsh and bash
> completions I started, and then got frustrated by how difficult it is to
> write bash and zsh completions....

FWIW my experience is that, while very cool and high "aaaah" factor,
bash completions greatly degrade the usability of bash. I never enable
them on my own systems. For example it breaks completion with the tar
command on annarchy -- I haven't figured out how to turn them off but
every time it's a huge pain that I can't tab-complete the name of a
tar archive because it doesn't know about .xz or something.

  -ilia