44941[BUGFIX] Allow overlaying of records for all languages again
Commit message
Review
-
Code review CR +2
- Oliver Hader +2
- Joschi Kuphal +1
-
Verified V +2
- Oliver Hader +2
- Joerg Kummer +1
Issues 1
Files 1
Review log 20
-
Mathias Schreiber uploaded patch set 1 11/25/15, 15:45
-
Mathias Brodala voted on patch set 1 CR +1 11/25/15, 15:49
-
Andreas Kienast commented on patch set 2 11/25/15, 15:51
Published edit on patch set 1 -
Joschi Kuphal commented on patch set 2 CR -1 11/25/15, 15:55
I don't think that simply reverting the change is a proper solution, as the change itself had its justification. Instead, the untouched / non-overlayed row should be returned (instead of swallowed). -
Mathias Brodala commented on patch set 2 CR -1 11/25/15, 16:02
You're right and I actually said the same before: all records with sys_language_uid=-1 should be returned untouched, no matter why they end up here in the first place. -
Mathias Schreiber commented on patch set 2 11/25/15, 16:10
The call stack is wrong. Simple as that. Do not overlay records that should not be overlayed - period. So the problem lies up higher in the call-stack. Fixing this is a huge undertaking which we will not do because we will change the way how we will handle translations in general - which would render this effort completely in vain. Please reconsider the -1 you set. -
Joschi Kuphal commented on patch set 2 11/25/15, 16:46
Sorry, but I don't agree. I do agree that the call stack is wrong, yes, and that the problem *should* be tackled on a higher level — which won't happen anytime soon for the given reasons. But in the meantime, instead of simply reverting the change and overlay *all* records again, you could at least improve the situation *a little* by only overlaying records with `sys_language_uid=0` and return each $row with `sys_language_uid=-1` just as is — which just doesn't happen (an never did). This solution wouldn't harm anybody and still fit the purpose of the original change. Every other solution would simply be unreasonable now that an action has to be taken anyway. -
Andreas Kienast commented on patch set 2 11/25/15, 16:56
imo we should revert the change at first to solve that regression. We can make that stuff cleaner at a later point -
Joschi Kuphal commented on patch set 2 11/25/15, 17:01
It's a two-liner to considerably improve the situation, without side effects. Why avoid this? -
Mathias Brodala uploaded patch set 3 11/26/15, 08:36
-
Mathias Brodala commented on patch set 3 11/26/15, 08:37
First stab at returning such records untouched. -
Joschi Kuphal commented on patch set 3 CR +1 11/26/15, 09:02
Looks good to me now! :) -
Mathias Schreiber commented on patch set 3 11/26/15, 10:40
I will re-push the first patchset again some time today. Reasoning is that this does not only affect languages but also workspace version overlays. As such, the impact of the changed behavior cannot be fully grasped at the moment, leading to potentially more problems than visible at first sight. -
Joerg Kummer commented on patch set 3 V +1 11/30/15, 21:07
It works. Did fix a launch today, thx. -
Markus Klein commented on patch set 3 11/30/15, 23:40
PageRepository.php:476ticket for the todo? -
Oliver Hader voted on patch set 3 CR +1 12/01/15, 08:40
-
Oliver Hader commented on patch set 3 12/01/15, 08:41
Looks good, first I was puzzled on the versionOL inside the "language === 0" block, but this always has to be called explicitly prior to getRecordOverlay. Fine then :) -
Oliver Hader voted on patch set 3 CR +2 V +2 12/01/15, 11:28
-
Oliver Hader commented on patch set 3 12/01/15, 11:28
-
Gerrit merged patch set 3 12/01/15, 11:47