switch to cronsim for cron parsing

[st31ny]

Avoid DST related bugs, such as those reported in #7.

CronSim is an alternative implementation of cron parsing in python and was (according to its documentation) explicitly designed to follow Debian's cron implementation, including the DST behavior.

My tests look quite promising and the issues I had noticed vanished using no tz parameter. Passing a zoneinfo timezone to the crontab decorator worked, as well as dateutil.tz.tzlocal().

Unfortunately, this is a breaking change since I had to remove croniter_kwargs, of course.

[pde-rent]

v2 upgrade with Cronsim broke our prod. They decided to not support the 6th optional expr (seconds) to not diverge from Posix's specs. We'll have to do with 1.8 and croniter for the time being

[st31ny]

v2 upgrade with Cronsim broke our prod. They decided to not support the 6th optional expr (seconds) to not diverge from Posix's specs. We'll have to do with 1.8 and croniter for the time being

True, didn't notice that beforehand. So probably we should try to get this into CronSim. Since seconds would simply add another field, I do not see a reason not to support seconds.

Maybe you could open a bug report at CronSim, @pde-rent ?

[pde-rent]

Maybe you could open a bug report at CronSim, @pde-rent ?

It seems like it was part of the project early on, but removed to remain posix compliant
here is the commit

Now if you were to bring this up again as integration bottleneck, maybe cronsim's contributors would re-consider?

[st31ny]

Now if you were to bring this up again as integration bottleneck, maybe cronsim's contributors would re-consider?

Sure, feel free to open the issue.