Re: [PATCH v3 14/16] iotests/linters: Add workaround for mypy bug #9852

[Hanna Reitz <hreitz@redhat.com>]

On 22.09.21 22:37, John Snow wrote:
>
>
> On Fri, Sep 17, 2021 at 7:16 AM Hanna Reitz <hreitz@redhat.com 
> <mailto:hreitz@redhat.com>> wrote:
>
>     On 16.09.21 06:09, John Snow wrote:
>     > This one is insidious: if you use the invocation
>     > "from {namespace} import {subpackage}" as mirror-top-perms does,
>     > mypy will fail on every-other invocation *if* the package being
>     > imported is a package.
>     >
>     > Now, I could just edit mirror-top-perms to avoid this
>     invocation, but
>     > since I tripped on a landmine, I might as well head it off at
>     the pass
>     > and make sure nobody else trips on the same landmine.
>     >
>     > It seems to have something to do with the order in which files are
>     > checked as well, meaning the random order in which set(os.listdir())
>     > produces the list of files to test will cause problems
>     intermittently.
>     >
>     > mypy >= 0.920 isn't released yet, so add this workaround for now.
>     >
>     > See also:
>     > https://github.com/python/mypy/issues/11010
>     <https://github.com/python/mypy/issues/11010>
>     > https://github.com/python/mypy/issues/9852
>     <https://github.com/python/mypy/issues/9852>
>     >
>     > Signed-off-by: John Snow <jsnow@redhat.com
>     <mailto:jsnow@redhat.com>>
>     > ---
>     >   tests/qemu-iotests/linters.py | 3 +++
>     >   1 file changed, 3 insertions(+)
>     >
>     > diff --git a/tests/qemu-iotests/linters.py
>     b/tests/qemu-iotests/linters.py
>     > index 4df062a973..9c97324e87 100755
>     > --- a/tests/qemu-iotests/linters.py
>     > +++ b/tests/qemu-iotests/linters.py
>     > @@ -100,6 +100,9 @@ def run_linters(
>     >                   '--warn-unused-ignores',
>     >                   '--no-implicit-reexport',
>     >                   '--namespace-packages',
>     > +                # Until we can use mypy >= 0.920, see
>     > +                # https://github.com/python/mypy/
>     <https://github.com/python/mypy/issues/9852>issues
>     <https://github.com/python/mypy/issues/9852>/9852
>     <https://github.com/python/mypy/issues/9852>
>     > +                '--no-incremental',
>     >                   filename,
>     >               ),
>
>     I’m afraid I still don’t really understand this, but I’m happy
>     with this
>     given as the reported workaround and you saying it works.
>
>     Reviewed-by: Hanna Reitz <hreitz@redhat.com
>     <mailto:hreitz@redhat.com>>
>
>     Question is, when “can we use” mypy >= 0.920?  Should we check the
>     version string and append this switch as required?
>
>
> The answer to that question depends on how the block maintainers feel 
> about what environments they expect this test to be runnable under. I 
> lightly teased kwolf once about an "ancient" version of pylint they 
> were running, but felt kind of bad about it in retrospect: the tests I 
> write should "just work" for everyone without them needing to really 
> know anything about python or setting up or managing their 
> dependencies, environments, etc.
>
> (1) We can use it the very moment it is released if you're OK with 
> running 'make check-dev' from the python/ directory. That script sets 
> up its own virtual environment and manages its own dependencies. If I 
> set a new minimum version, it will always use it. (Same story for 
> 'make check-tox', or 'make check-pipenv'. The differences between the 
> tests are primarily on what other dependencies they have on your 
> environment -- the details are boring, see python/Makefile for further 
> reading if desired.)
>
> (2) Otherwise, it depends on how people feel about being able to run 
> this test directly from iotests and what versions of mypy/pylint they 
> are using. Fedora 33 for instance has 0.782-2.fc33 still, so I can't 
> really "expect" people to have a bleeding-edge version of mypy unless 
> they went out of their way to install one themselves. (pip3 install 
> --user --upgrade mypy, by the way.) Since people are used to running 
> these python scripts *outside* of a managed environment (using their 
> OS packages directly), I have largely made every effort to support 
> versions as old as I reasonably can -- to avoid disruption whenever I 
> possibly can.
>
> So, basically, it kind of depends on if you want to keep 297 or not. 
> Keeping it implies some additional cost for the sake of maximizing 
> compatibility. If we ditch it, you can let the scripts in ./python do 
> their thing and set up their own environments to run tests that should 
> probably "just work" for everyone.297 could even just be updated to a 
> bash script that just hopped over to ./python and ran a special 
> avocado command that ran /only/ the iotest linters, if you wanted. I 
> just felt that step #1 was to change as little as possible, prove the 
> new approach worked, and then when folks were comfortable with it drop 
> the old approach.

Hm.  So the gist is, since we apparently want to keep 297, we have to 
wait for some indefinite time until in some years or so someone 
remembers this workaround and removes it?

Doesn’t sound quite ideal, but well...

Hanna