[bug#39777,V3,02/11] gnu: python-cachy: Update to 0.3.0.

* gnu/packages/python-xyz.scm (python-cachy): Update to 0.3.0.
[arguments]: Disable failing tests.
---
 gnu/packages/python-xyz.scm | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

On Mon, Mar 09, 2020 at 09:18:18AM +0100, Tanguy Le Carrour wrote:
> * gnu/packages/python-xyz.scm (python-cachy): Update to 0.3.0.
> [arguments]: Disable failing tests.

> +    (arguments
> +     '(#:phases
> +       (modify-phases %standard-phases
> +         (add-before 'check 'disable-failing-tests
> +           (lambda _
> +             (substitute* "tests/stores/test_memcached_store.py"
> +               (("def test_forever") "def _test_forever")
> +               (("def test_increment") "def _test_increment")
> +               (("def test_decrement") "def _test_decrement")
> +               (("def test_put_numeric_value") "def _test_put_numeric_value")
> +               (("def test_put_value_into_memcache") "def _test_put_value_into_memcache")
> +               (("def test_value_is_returned") "def _test_value_is_returned")
> +               (("def test_value_is_returned_for_numerics") "def _test_value_is_returned_for_numerics"))
> +             #t)))))

Do you have any idea what's going on with these tests?

We should at least add a comment here, explaining why skipping the tests
is not hiding something that will be a problem while using cachy.

Le 03/11, Leo Famulari a écrit :
> On Mon, Mar 09, 2020 at 09:18:18AM +0100, Tanguy Le Carrour wrote:
> > * gnu/packages/python-xyz.scm (python-cachy): Update to 0.3.0.
> > [arguments]: Disable failing tests.
> 
> > +    (arguments
> > +     '(#:phases
> > +       (modify-phases %standard-phases
> > +         (add-before 'check 'disable-failing-tests
> > +           (lambda _
> > +             (substitute* "tests/stores/test_memcached_store.py"
> > +               (("def test_forever") "def _test_forever")
> > +               (("def test_increment") "def _test_increment")
> > +               (("def test_decrement") "def _test_decrement")
> > +               (("def test_put_numeric_value") "def _test_put_numeric_value")
> > +               (("def test_put_value_into_memcache") "def _test_put_value_into_memcache")
> > +               (("def test_value_is_returned") "def _test_value_is_returned")
> > +               (("def test_value_is_returned_for_numerics") "def _test_value_is_returned_for_numerics"))
> > +             #t)))))
> 
> Do you have any idea what's going on with these tests?
> 
> We should at least add a comment here, explaining why skipping the tests
> is not hiding something that will be a problem while using cachy.

Those ones are out of my comfort zone! :-(

Seems that all the `RedisStoreTestCase` are failing with messages like:

```
ResourceWarning: unclosed <socket.socket fd=3, family=AddressFamily.AF_INET, type=SocketKind.SOCK_STREAM, proto=0, laddr=('127.0.0.1', 33754)>
  if self._get_socket():
ResourceWarning: Enable tracemalloc to get the object allocation traceback
```

Might be related to the Redis server that is supposed to be running
during the tests?!

Again, any idea welcome!

On Thu, Mar 12, 2020 at 09:58:22AM +0100, Tanguy Le Carrour wrote:
> Seems that all the `RedisStoreTestCase` are failing with messages like:
> 
> ```
> ResourceWarning: unclosed <socket.socket fd=3, family=AddressFamily.AF_INET, type=SocketKind.SOCK_STREAM, proto=0, laddr=('127.0.0.1', 33754)>
>   if self._get_socket():
> ResourceWarning: Enable tracemalloc to get the object allocation traceback
> ```
> 
> Might be related to the Redis server that is supposed to be running
> during the tests?!

Okay, networking is not possible in the build environment, so for the
tests that depend on connecting to a server, you can add a comment like
"These tests require networking.".

Is that all of them?

On Thu, Mar 12, 2020 at 09:58:22AM +0100, Tanguy Le Carrour wrote:
> Seems that all the `RedisStoreTestCase` are failing with messages like:
> 
> ```
> ResourceWarning: unclosed <socket.socket fd=3, family=AddressFamily.AF_INET, type=SocketKind.SOCK_STREAM, proto=0, laddr=('127.0.0.1', 33754)>
>   if self._get_socket():
> ResourceWarning: Enable tracemalloc to get the object allocation traceback
> ```

I just got word that networking to localhost *is* possible in the build
environment. So it's not as simple as that in this case. Can you check
if the Redis server is running?

Le 03/14, Leo Famulari a écrit :
> On Thu, Mar 12, 2020 at 09:58:22AM +0100, Tanguy Le Carrour wrote:
> > Seems that all the `RedisStoreTestCase` are failing with messages like:
> > 
> > ```
> > ResourceWarning: unclosed <socket.socket fd=3, family=AddressFamily.AF_INET, type=SocketKind.SOCK_STREAM, proto=0, laddr=('127.0.0.1', 33754)>
> >   if self._get_socket():
> > ResourceWarning: Enable tracemalloc to get the object allocation traceback
> > ```
> > 
> > Might be related to the Redis server that is supposed to be running
> > during the tests?!
> 
> Okay, networking is not possible in the build environment, so for the
> tests that depend on connecting to a server, you can add a comment like
> "These tests require networking.".

Good to know, thanks!


> Is that all of them?

I guess so! I'll double check, update the patch and submit it!

Thanks!

Le 03/14, Leo Famulari a écrit :
> On Thu, Mar 12, 2020 at 09:58:22AM +0100, Tanguy Le Carrour wrote:
> > Seems that all the `RedisStoreTestCase` are failing with messages like:
> > 
> > ```
> > ResourceWarning: unclosed <socket.socket fd=3, family=AddressFamily.AF_INET, type=SocketKind.SOCK_STREAM, proto=0, laddr=('127.0.0.1', 33754)>
> >   if self._get_socket():
> > ResourceWarning: Enable tracemalloc to get the object allocation traceback
> > ```
> 
> I just got word that networking to localhost *is* possible in the build
> environment. So it's not as simple as that in this case. Can you check
> if the Redis server is running?

*ARF* should have read all of it! :-(

I'll investigate! Thanks for the advice.

Hi Leo,

Le 03/14, Leo Famulari a écrit :
> On Thu, Mar 12, 2020 at 09:58:22AM +0100, Tanguy Le Carrour wrote:
> > Seems that all the `RedisStoreTestCase` are failing with messages like:
> > 
> > ```
> > ResourceWarning: unclosed <socket.socket fd=3, family=AddressFamily.AF_INET, type=SocketKind.SOCK_STREAM, proto=0, laddr=('127.0.0.1', 33754)>
> >   if self._get_socket():
> > ResourceWarning: Enable tracemalloc to get the object allocation traceback
> > ```
> 
> I just got word that networking to localhost *is* possible in the build
> environment. So it's not as simple as that in this case. Can you check
> if the Redis server is running?

I checked and this bug has been reported upstream last year and is not
yet fixed.
There's a workaround, but it requires to package `pifpaf` and 5 more
dependencies. I'll eventually do that… later next week!?

In the meantime, I'm pushing v4 of the patch set if you're OK with
disabling the failing tests and adding a comment pointing to the bug report.
If you're not, which is perfectly fine, I've reorganized the stack so that
you could commit the 8 patches up to `python-cachy`.

Regards

On Thu, Mar 19, 2020 at 06:27:05PM +0100, Tanguy Le Carrour wrote:
> In the meantime, I'm pushing v4 of the patch set if you're OK with
> disabling the failing tests and adding a comment pointing to the bug report.
> If you're not, which is perfectly fine, I've reorganized the stack so that
> you could commit the 8 patches up to `python-cachy`.

That sounds fine, thanks!