No need to be default ...

Capture d’écran du 2026-05-28 11-12-11.png (83,346 bytes)   

Capture d’écran du 2026-05-28 11-12-11.png (83,346 bytes)   

Confirmed: updating skelvanilla and extend_skelvanilla with <version>7</version> fixes the issue.

(But please, don't do it automatically.)

Is there something to fix then or not?

Is there something to fix then or not?

Clearly yes: a simple user cannot update and gets stuck on the update screen.

:( extend fruity work: one time and it's OK...

Sorry didn't get what you said in your last comment. Can you clarify?

I test with a new installation

1. Set up v6.x
2. Extend vanilla
3. Create survey and set extend_vanilla as theme
4. Update

It's OK :(

There are something other in my previous test. BUT i didn't keep the database .....

Got it, please let me know if you figure it out or if I can close this ticket then.

I try to make more test.
No reason it happen once (see screencast) a,nd not again.

Else : i close it for "No reproduce" :(

PR for the issue can be found at https://github.com/LimeSurvey/LimeSurvey/pull/5015

Arpad made a PR https://github.com/LimeSurvey/LimeSurvey/pull/5015. Please take a look Denis.

Arpad made a PR https://github.com/LimeSurvey/LimeSurvey/pull/5015. Please take a look Denis.

OK ! Seems OK, i check with Group setting to skelvanilla then.

Then need

1. Check all installed themes: check if they are compatible with 7, uninstall if not
2. If no installed theme : install defaultfixedtheme
3. If defaultfixedtheme is not compatible, install fruity_twentythree
4. For each installed theme : check extended not compatible in SurveyGroup
5. My opinion : set it to 'inherit' , Arpad set it to parent
6. Finally check Group 0 theme, and set it to the defaultfixedtheme or the last theme
7. Same for defaulttheme, and set it to the defaultfixedtheme or the last theme

ping @lajosarpad

PS : we can set defaultfixedtheme to fruity_twentythree in config-default now.

We can set defaultfixedtheme to fruity_twentythree

Also, please feel free to implement any suggestion that you may have and ping me here to reread the PR.

Here, you do it only for fruity_twentythree : right ?

Not for extended vanilla, bootstrap, etc. OR skelvanilla and others.

The PR is setting the template to fruity_twentythree if it's not either fruity_twentyhtree, an extension or inherit.

The PR is setting the template to fruity_twentythree if it's not either fruity_twentyhtree, an extension or inherit.

WHAT ???

1. extends_fruity_twentythree just before update: uninstalled since not compatible (config.xml show <vesion>6</version> only)
2. We have vanilla + fruity + bootswatch in core
3. skelvanilla is compatible with 7 now (and I think we'll have more in the future, since 6 versions are directly compatible with 7).

Okay, so your position is that we do have some themes that are not fruity_twentythree which are compatible with version 7 which should not be changed upon upgrade. And hence, if I understand you properly, then the proposal would be to check which templates are compatible with LS7 and update those that aren't. Am I understanding you accurately?

we do have some themes that are not fruity_twentythree

vanilla + fruity + bootswatch are already in core ! https://github.com/LimeSurvey/LimeSurvey/tree/master/themes/survey
What do you say herre ?

check which templates are compatible with LS7

If someone creates a 6+7 compatible template, why do we update it to inherit?

check which templates are compatible with LS7

Why disable the vanilla core theme but allow extend_fruity_twentyhtree not compatible with 7?

"If someone creates a 6+7 compatible template, why do we update it to inherit?"

We are not updating anything to inherit. We exempt inherit from being updated.

"Why disable the vanilla core theme but allow extend_fruity_twentyhtree not compatible with 7?"

My understanding was that fruity_twentythree was to be supported along with extensions and nothing else. If this understanding was wrong, then the PR needs to be adjusted.

We are not updating anything to inherit. We exempt inherit from being updated.

You say "The PR is setting the template to fruity_twentythree if it's not either fruity_twentyhtree, an extension or inherit."

And here, looking at code : if User set default group to vanilla : the code move it fruity_twentythree

My understanding was that fruity_twentythree was to be supported along with extensions and nothing else. If this understanding was wrong, then the PR needs to be adjusted.

Where did you see that? I never write this!
I just say, I have a loop when I try to upgrade with a non-compatible theme.

"And here, looking at code : if User set default group to vanilla : the code move it fruity_twentythree"

As explained in my previous comment, if a template is not fruity_twentythree nor an extension of it, then the PR sets it to fruity_twentythree. So this is what the PR is doing:

1. template is "fruity_twentythree" -> do nothing
2. template is an extension of "fruity_twentythree" -> do nothing
3. template is "inherit" -> do nothing
4. template is something else -> changes it to fruity_twentythree

This is the current behavior of the PR. You expressed disagreement, so I said the following:

"My understanding was that fruity_twentythree was to be supported along with extensions and nothing else. If this understanding was wrong, then the PR needs to be adjusted."

which of course not something you said, because it was me who said it. I was trying to tell you that if the behavior of the PR is incorrect, then it is due to a misunderstanding of the intentions on my part.

OK, there are isse, but not real loop

I extend vanilla to extend_vanilla , adn set it as Default theme of group
I extend fruity_twentythree to extend_fruity_twentythree , and set it as Default theme of group 0 and a new group and as default

I have 2 times the alert before the update + one time after the update for the default theme.

Last check, and I close the issue.

Thank you. I can close the PR then.

Please wait,

Did you see the screencast ?
There is a potential loop here!

I wite «Last check, and I close the issue.»

... there are a reason.

I think I'm old enough to close the issue on my own.

No loop again, just 2 time (related to the 2 group ? )