cron #previous_time loop breaker issue

[bdarcet]

cron #previous_time loop breaker issue

[jmettraux]

Hello,

I have further refined how et-orbi computes rweek which is used for the cron modulo. I think it solves your issue.

Please upgrade to et-orbi 1.2.10 and tell me how it goes.

Merci beaucoup !

[bdarcet]

Hello, @jmettraux !
Thanks again for your fast reply on this issue :)
It looks like the upgrade to et-orbi 1.2.10 fixed all our errors 🎉

I've done a bit of testing to compare our last used versions 1.2.7 & 1.2.10
And it looks like the results are a bit different between the 2 (not for every cron but for some)

I just wanted your opinion on those (it looks like on 1.2.7, we had big jumps in the future on the next_time for example that we don't have anymore - I'm not sure if they were expected in the first place but .. - etc):

Version	1.2.7	1.2.10
cron	--- 0 5 * * 2%6 Europe/Berlin ---	--- 0 5 * * 2%6 Europe/Berlin ---
previous time	2024-02-27 04:00:00 +00:00 Tuesday	2024-02-27 04:00:00 +00:00 Tuesday
next time	2024-11-05 04:00:00 +00:00 Tuesday	2024-04-16 03:00:00 +00:00 Tuesday

Version	1.2.7	1.2.10
cron	--- 0 10 * * 2%2 America/New_York ---	--- 0 10 * * 2%2 America/New_York ---
previous time	2024-03-12 14:00:00 +00:00 Tuesday	2024-03-19 14:00:00 +00:00 Tuesday
next time	2024-11-05 15:00:00 +00:00 Tuesday	2024-04-02 14:00:00 +00:00 Tuesday

Version	1.2.7	1.2.10
cron	--- 20 0 * * 2%2 America/Los_Angeles ---	--- 20 0 * * 2%2 America/Los_Angeles ---
previous time	2024-02-27 08:20:00 +00:00 Tuesday	2024-03-19 07:20:00 +00:00 Tuesday
next time	2024-04-02 07:20:00 +00:00 Tuesday	2024-04-02 07:20:00 +00:00 Tuesday

Version	1.2.7	1.2.10
cron	--- 20 9 * * 2%8 America/New_York ---	--- 20 9 * * 2%8 America/New_York ---
previous time	2024-03-12 13:20:00 +00:00 Tuesday	2024-03-19 13:20:00 +00:00 Tuesday
next time	2024-12-17 14:20:00 +00:00 Tuesday	2024-05-14 13:20:00 +00:00 Tuesday

Let me know, thank you!

[jmettraux]

Hello @bdarcet,

thanks for the extra test cases. I much prefer the new 1.2.10 results. I've added some debug output to watch how #next_time and #previous_time zero on their target and it seems correct to me, it seems it does not skip now (it skipped too much and triggered the loop breaker previously, hence your reports).

The initial implementation of #rweek was too loose, sorry about that. I'm glad you spotted the issue.

Closing this one now. Please report if there is anything else.

Have a nice week-end!