Spring board for known (+ custom) web applications

[MichaIng]

Since we have some users who want to use something like Heimdall, to have a page with tiles/links to various web applications, but Heimdall is difficult to implement into dietpi-software in a compatible way, probably we can implement it into DietPi-Dashboard:

• We could scan for installed (dietpi-)software titles where we know the subpath or the port, and show a related tile with link (using the dashboard access host + known default port and/or subpath) and a hardcoded logo.
• We could allow to edit the link, probably even the logo.
• We could allow to add custom tiles with link and logo.
• We could allow to reorder the tiles.

The last three parts would require a server-side storing of the info, so something which DietPi-Dashboard does not do at all. Of course it could be also done just via config.toml for now, so that it is possible to override the default link of a defined web application.

[TommySharpNZ]

I look forward to this functionality :-)

[nonnorm]

Personally, I still think it would be better to have a dedicated bookmarks dashboard application, however this detecting installed software from DietPi-Software would be nice. It would probably require putting the ports and web addresses of software in a more standardized place, though. Not sure about how to do server-side storage, a full database seems like too much, and I like the no dependencies. Maybe storing it in a JSON file, similar to Python's pickledb. It would also allow graph data to be saved, like in #153.

[MichaIng]

Earlier or later, adding the ability to change config.toml settings via web interface would be great anyway. It could then also hold overrides for the known/default software ports and/or URLs (when running on a different machine or requiring a different host than the dashboard). Of course it would be also a dedicated config file if all in one is seen as an issue, not sure which format is best then, I personally don't like JSON because it is hard to edit manually and hard to grep/sed/awk. But of course there are processing libraries and tools available. For custom entries I however agree that TOML is not great either, requiring either custom sections or variable names with incrementing suffix or such, like spring_url_0, spring_url_1, spring_label_0, spring_color_0 or so (just examples of possibly interesting customisations).

[MichaIng]

A vote for this from the forum: https://dietpi.com/phpbb/viewtopic.php?t=10366

[nonnorm]

Before any work can be done on this, it would be great to have some kind of port database for DietPi-Software. Otherwise, users would need to add the software themselves (it seems like automatically getting the ports would be a big advantage over Heimdall). Pickledb now has feature flags to remove unneeded dependencies, so customizing and storing on the server could be done.

[MichaIng]

ss -tlpn could be used to get process names which listen on specific TCP ports. Probably there is a Rust crate as well. Otherwise, I agree, a port database makes much sense, also to see port conflicts reliably when impending software. However, reasonably ports can always be changed, so probably it is even best to derive available software and related ports directly from ss, so we only need a list of known process names, and probably associated logos?

Ah, and then there are the port 80/443 webserver applications and their sub path. Okay probably better to parse /boot/dietpi/.installed then, as those cannot be detected in a generic way otherwise.

[nonnorm]

Currently that sounds like a good plan. It would probably not be stored as a markdown file, perhaps as JSON or YAML, which makes it easiest to write from the frontend.

[MichaIng]

Keeping the final implementation in mind, at least I wouldn't implement something "smaller" now which needs to be completely rewritten later, but only if the code can be easily extended for the final implementation. Just my 2 cents, to reduce coding overhead 😉.

[nonnorm]

This should be extendable, it would probably be how you would manually add software that's not automatically detected, and how it would be stored in the backend anyway.

[nonnorm]

I'll begin work on this as soon as I get #232 and #236 merged.

[TommySharpNZ]

Hi Team, any updates on this? I'd love to get rid of my docker setup that just hosts Heimdall... :-)