Warning count from 1932 to 1842

[cbeauchesne]

See #1133 for more details

[cbeauchesne]

@tsauerwein by chance, would it possible to have your opinion about this change : https://github.com/c2corg/v6_api/pull/1134/files#diff-bcdcacbeeac9f27f3ad65cda8f9decaee8aa8358a2cc6bf4709d821a2ccd637aR82

I added it because we have a warning saying The Redis backend thread_local_lock parameter should be set to False when distributed_lock is True : https://github.com/sqlalchemy/dogpile.cache/blob/master/dogpile/cache/backends/redis.py#L108

But I have absolutely no idea of its purpose, and it scares me a little bit :)

Do you know if there any side-effect with this parameter ?

[lbesson]

I think that comes from one of the (outdated) dependencies. We already have this warning in the master branch and that still works.

[tsauerwein]

@tsauerwein by chance, would it possible to have your opinion about this change : https://github.com/c2corg/v6_api/pull/1134/files#diff-bcdcacbeeac9f27f3ad65cda8f9decaee8aa8358a2cc6bf4709d821a2ccd637aR82

I added it because we have a warning saying The Redis backend thread_local_lock parameter should be set to False when distributed_lock is True : https://github.com/sqlalchemy/dogpile.cache/blob/master/dogpile/cache/backends/redis.py#L108

But I have absolutely no idea of its purpose, and it scares me a little bit :)

Do you know if there any side-effect with this parameter ?

Checking the dogpile.cache documentation, your change makes sense. The configuration uses distributed_lock=True because the API runs in multiple processes/containers/servers and the lock should be synchronized across all. As I understand it, distributed_lock and thread_local_lock are mutually exclusive. So it would make sense to set thread_local_lock to False.

[lbesson]

don't want to update bbcode while you're at it? :D

[cbeauchesne]

@lbesson I've a loooooong todo list on API. I'm focusing on the easy one, and on ones that makes dev's life easier, sort of getting a foot in the door in a way :)