What's coming in 7.0 (was 6.4)
-
@girish said in What's coming in 6.4:
I added a last item "Make it easy to install non-app store apps".
That is great news. In case you do not yet have a better idea how this could be designed, I want to pitch the way this is organised in Portainer again. I explained it already a while back in https://forum.cloudron.io/topic/4485/proposal-the-cur-cloudron-user-repository/14?_=1626194230000. Basically a local admin can override/extend the entries from their official "appstore". Additional entries are defined in a json structure. When opening the Cloudron appstore it makes a request to
https://my.cloudron.host/api/v1/appstore/appswhich gets a json response with all the appstore data. What if opening the store would in addition to this also make a request tohttps://user:password@store.my.domain/apps.jsonand also show a category "community" that lists apps from this json listing. Plus points if the listing can be password protected (like the private docker registry). This could also be a nice revenue model for external app developers. -
G girish pinned this topic on
-
@girish said in What's coming in 6.4:
Make it easy to install non-appstore apps.
Can someone point me to any prior discussions based around this topic?
Even though I myself like to use these non-appstore apps, partly to help test, partly because it is an app I'd like to use (I just installed the Paperless-ng app), the amount of troubleshooting that goes on seems like alot. And, as I've expressed before in other posts, I love the stability and ease of Cloudron, and the amazingly quick and attentive help the main devs give, particularly on this forum.
So I don't see how adding the option for more people, likely many who are even less skilled than I (and I am barely keeping my head above water here!), to start messing around with apps that need tweaking, is going to help keep all that we love about Cloudron at the fore. I mean, the tech barrier (code barrier? comprehension barrier?) in front of building and installing non-app store apps I imagine acts as a gate to limit the number of Help requests connected to these apps. Do we need another type of gate?
I am impressed by how quickly and neighbourly the other posters are to help, both myself, and I can see them helping others. Is there not some other way to keep this third-party option alive, maybe even another subscroption level?
Sub-reddits often require posters to have a certain life-span before you can post; lowendtalk also has a minimum activity level for posting. But I can see how that could seem too exclusive for new comers, and maybe even for old-timers now.
Any thoughts?
-
@scooke I think I’m with you on that. Having ‚inofficial‘ apps or a second AppStore doesn’t help quality on the long run - unless it’s just testing ground for officially supported apps. Otherwise it’s easy to habe the Nextcloud or Yunohost mess on your hand with unofficial apps breaking all sorts of stuff….
-
@scooke Yeah, great points. I should have been more clear. We don't intend to add a way to add a 3rd party app store. While this is possible, apart from it being some technical achievement, will most likely just frustrate users.
What I meant by that feature was that it will be useful to have a way to quckly install apps that are packaged by others without having to step into the CLI. i.e skip the whole CLI install+build+push image+install cycle. This workflow requires the user to know nodejs, docker, CLI use among other things, quite complicated. We also have a selfish reason to do this. When someone suggests that something is "packaged", we would like to have a quick look as to what state the app is in and we have to do this CLI workflow ourselves.
I don't know how this looks like but maybe there is some simple install UI where one can just put upload a manifest file + docker image name and that's it.
-
Yes, depending on quality this could of course cause some confusion. But there are a few examples of working third party repositories out there that extend a main product. The external app stores on Synology come to mind for example. Plus since apps are running in read-only containers and are mostly isolated from the host I don't think an app can mess up a server at all. But the most important part for me was in the last sentence:
@fbartels said in What's coming in 6.4:
This could also be a nice revenue model for external app developers.
-
@girish said in What's coming in 6.4:
We also have a selfish reason to do this. When someone suggests that something is "packaged", we would like to have a quick look as to what state the app is in and we have to do this CLI workflow ourselves.
This makes a lot of sense actually. With whatever process you come up with, the goal would be that knowing when someone says the app is packaged, "ready", it most likely is because they've had to do the correct steps, get auto-checked. So in the long run it would actually help the AppStore by freeing up your time checking things before making them Official.
I imagine there'll still be alot of troubleshooting involved getting to the Packaged state though.
-
@girish thank you! Any way to add configurable backup failure SSL expiration notifications? I liked it very much the way there were before 6.3, as I only have one backup per day and 3 missing backups - it’s 3 days of data loss.
For SSL - I have a non-standard configuration and having notifications in advance helps really a lot. -
Excellent progress!
-
@girish I've got a suggestion if the focus on this release is on email : add some basic feature to deal with mailing-lists:
- add at least an API to see the members of a mailing-list. (priority P3)
- add an API to subscribe/unsubscribe a user from a mailing-list. (priority P2)
- build an UI for both features (P1)
In our use case, a user sending an email to a mailing-list want to be sure that someone from the crew is receiving the mail, aka is member of the mailing-list.
I definitely don't want to go through the hassle of installing/configuring mailman or something else, because I feel that the cloudron mailing-list feature does just enough for the job, and just need a little bit more basic features.
It would be very nice if a user can subscribe and unsubscribe by himself for the mailing-list, taking the burden off the administrator for adding/removing manually users. Besides, it would give more autonomy to the users, which is good.
With at least some basic APIs, I could automate myself the process, through a rocketchat bot for instance.
-
@girish oh, one more thing, fix the "https://forum.cloudron.io/topic/2611/cannot-send-email-from-outlook-2007-with-5-2-4-connection-error-ssl-routines-tls/2" instead of having a workaround...
-
@samir said in What's coming in 6.4:
fix the "https://forum.cloudron.io/topic/2611/cannot-send-email-from-outlook-2007-with-5-2-4-connection-error-ssl-routines-tls/2" instead of having a workaround...
The problem here is outlook that uses old encryption by default. Changing it would weaken the security of every other mail client.
-
@potemkin_ai About backups, let me look into if there is a better approach.
Can you tell me a bit more about your SSL setup? Note that cert renewal failure notifications are still there, they have not been removed. It's just that it won't alert you 30 days in advance now and instead only 10 days in advance (but it starts renewing 30 days in advance). This allows for 20 days of let's encrypt to be flaky.
-
@samir The current mailing list on Cloudron isn't actually a traditional mailing list i.e one with subscribe/unsubscribe feature. It is really just a forwarding list. Meaning if a mail comes to a specific address, it forwards it to unconditionally to all the members.
That said, there is already an API to see members and add/remove people from the forwarding list. I will put it in the docs and link it here.
-
-
To give an update here, we sidetracked a bit and decided to modernize our codebase. Specifically, we moved from callback based programming to more modern async/await. Much of that work here is done, so we will post updates on features as we implement them.
