The first release candidate for NC 29 was published this week. By default, existing apps are not compatible with the new version and must be tested (and perhaps fixed). The goal is to get most of the popular apps with NC29 support until the release date of NC 29 (April 24th).
You can try to enable an app by clicking enable manually in the app overview (as an admin). Please leave a feedback on the bug tracker of an app if your test was successful (for follow-ups, leave a ) or leave a report with the errors you encountered.
Here is an auto-generated list of apps that were compatible with previous versions of Nextcloud but are not supported for NC 29. This list is contineously updated:
App ID
NC 27
NC 28
NC 29
Issue Tracker
aaochat
backup
bbb
chores
co2clock
diary
discoursesso
dokuwiki
emlviewer
extract
files_fulltextsearch_tesseract
files_markdown
files_mindmap
files_texteditor
gestion
googledocsredirect
groupfolder_filesystem_snapshots
hibp_login_check
integration_ews
integration_forgejo
integration_gptzero
integration_libretranslate
integration_pexip
integration_twitter
jitsi
journal
l10n_override
listman
llm
mail_roundcube
majordomo
nextcloudamarillo
nextframe
occweb
pride
quickaccesssorting
registertocontact
rides
salattime
sciencemesh
sensorlogger
sharemover
social
socialsharing_facebook
solid
souvenirs
transfer
tuya_cloud
twofactor_email
twofactor_reminder
user_ispconfig
user_pwauth
user_vo
vdirsyncerui
video_converter
Apps with testing versions:
App ID
version
Bugtracker
news
25.0.0-alpha8
files_archive
1.2.4-rc7
scimserviceprovider
1.0.0-alpha.1
pdf_downloader
1.2.3-rc3
Table updated on September 17 2024, 22:30 UTC
There are already 273 apps for NC 29
If breakage start because one leaves out stuff… maybe it is time to introduce some “NC25-Compat” kit “NC26-Compat” to at least provide the scaffolding of “ancient” stuff and allowing some transition?
It might make sense to have such a kit show there is a transition package active?..
So bootstrap is not just scrapped but phased out under control of user?
Nextcloud moves relatively fast with major versions (around 3 per year), but they try to keep the system for apps stable. For NC 27 however were quite a few changes and it needed some time to adjust most of the apps.
If you wanted to keep things compatible, even the php versions supported in NC 25 are partly not supported any more and vice versa.
If you have ideas, how to better resolve that, please open a dedicated topic for that to discuss that. I think there already was a topic where developers raised a few issues that made their work hard, but I couldn’t find it again.
Here we just try to keep track for the apps concerning the current release, so users/testers can report issues and follow the status.
Final note from me:
There is more stuff broken and fallen into the gutter beside the path NC chose.
I think more apps are broken in the app store than working.
I have no definitive idea about how to prevent this from reoccurring.
One idea: Ensure the Documentation (valid for the API) is published AHEAD of beta releases so developers can adjust.
(That also would imply there is some actual engineering going into NC).
Currently the practice seems to be throwing some Beta release over the wall, see what fallout it creates, fix that then describe remaining stuff into docs …
And hope the previous creators of APPS do follow…
(The release notes part, the API docs is a hotch potch of NC15 examples, NC17 Code, intersperced with NC11 descriptions… etc. trailing the current release with at least about a year worth of updates).
I just now wonder how long VUE will be used, 1, 2 or 3 versions before it gets replaced with the next whimp.
I’ll continue to use what works for now. and let go of some functionality if the next app falls down the side. Now it is pico_cms, before : pdfreader, files_opds etc.
I do develop code, just No web apps, no PHP, no Vue etc. I tried 2 years ago to revive some App only to find the state of mess called the NC X API which only loosely correlates to NC X+1 or X-1 API’s, less so to X-2 or X+2 API’s.
Learning everything is beyond my available time, and obsolete before i get enough grips on it.
On closing note: Maybe it would help to have NC developers assist App developers 1:1, it might also help the NC developers see HOW people actually use the platform…
and for 25 this post was in April for the release being published in October (and the first beta in August 2022).
They recently added new tutorials for future app developers:
In the forum, we have a section for app developers, if there are specific question, remarks. I can’t tell you for what reason which feature/function was removed or changed. And/or check with other developers how they do it.
With ahead i mean the API for 30 should be fixed by now, so any developer can count on it’s contract it provides for apps. It would allow to prepare for future changes.
Another approach … look at the API’s of Linux, OpenVMS syscalls stable for 30+, 50+ years… while all functions it provides did radically change in that period, got extended, and ported across all kind of platforms. Those API’s are rigid yet allow changes & growth.
Imagine Linus Torvalds radically change the API every two releases with invalidating anything for over 3 versions ago… (those are the worlds i come from).
Ok… Issue with cms_pico seems to mostly be the use of ILogger,
according to PSR-3 ( PSR — Nextcloud latest Developer Manual latest documentation ) this is the intended interface…
Just not available in NC29, i might be reading it wrong, in that case there is abiguity.
If such an update causes code breakage should it be in PSR-3? or a newer PSR?.
I’ll try to give the pull request a spin… i know it will break version verification from then on… :-(, mind you i am not a web developer, and due to the state of nextcloud between versions 20-25 i gave up on the idea of modifying apps.
Tried it.
NC29: next error in debug:
[cms_pico] Debug: /appinfo/app.php is deprecated, use \OCP\AppFramework\Bootstrap\IBootstrap on the application class instead.
GET /apps/logreader/api/poll?lastReqId=eVXRM7carbBsNv2ICrov