I'm Miles, also see: http://github.com/mil
My nick is mla on IRC.
REPORTED RESOLVED FIXED
-nocursor on X / from xdm is the biggest blocker on this.
I don't have too much experience but I remember testing unclutter originally and it was fairly unreliable in contrast to X -nocursor. But more testing would be needed here. Thanks for suggesting ~marderbot.
That said, aesthetically I'd be happy to just knock out the -nocursor altogether and have the cursor shown (that was how it was in Sxmo 0.1). I think the cursor is actually helpful to show where you last clicked. But i'm sure I'm in the minority in wanting the cursor back :)
I'm skeptical of switching to pipewire, if someone want's to take up the work and can demonstrate it works reliably.. ok but ALSA + dmix + dsnoop is a nice reliable solution. That said I don't use bluetooth.
Done, reverted back to the old reliable & fast idiotbox way https://lists.sr.ht/~mil/sxmo-devel/%3C20210610004718.11772-1-m%40milesalan.com%3E
I don't think we need to overcomplicate it idiotbox works - ytdl hardly does. I'd rather things work reliably then be slow or not work / add hooks / catch cases.
Eh i'm not sure, i do kinda like the quirkiness of the weather script, that said for sure Europe would be nice. Wttr is basically a 1-line curl thing.
One idea would be to integrate dmenu as a dock window that sits above svkbd rather then being an arbitrary floating window, overlaying the background windows via the center patch as is done currently. E.g. dmenu instead could automatically adjust WM space as svkbd does currently and pop up right above svkbd if svkbd is also visible at the time (or just dock to the bottom alone if not). I had a POC on this at one point but it may require some further refactoring on the dwm dock patch, but would be quite doable with minor changes on dwm side.
The other idea is "fullscreen" sans svkbd per your example ~fjc , Though I would say the top bar shouldn't be covered. The downside is that you loose the context of the window itself whereas with a dock-patch style you still can see the input window aside.
Added a deprecation notice to http://images.lrdu.org/ and updated install guide to point to bpo.
We might want to change out the link on all the pages that links to images too.
In principal I think having BPO generate images and deprecating (and even removing images.lrdu.org) is probably the right move. This provides consistency since (I think) we are the only UI to generate our own images and this is more of a 'distro' task.
The only things we'd loose two things by going with BPO fully are:
Weekly duration - historically I had published new Sxmo images on the (day) we tagged the release number in git / had things accepted upstream for APKBUILDs.
Historical images - it seems like BPO only has some history so we loose the timeline.
For (1), I don't think that's a major issue as image users are likely to be "end-consumers" and can wait a little while, if they want things up to date they can always
apk update/upgrade. For (2), I kind of liked the ability to go back in time to track regressions on an image basis, but I don't think it's entirely necessary.
I think I'll take a todo to put a forward on images.lrdu.org to images.postmarketos.org and update the docs for the new username password unless anyone has any objections (~ollieparanoid let me know if putting this forward in and going this way is good by you).
Sounds cool and will be excited to check out the svkbd changes soon. I think if things are done in suckless-compatible way, they won't have an issue upstreaming.
Vibration on tap sounds awesome - I wonder if this could just be a CLI flag like svkbd -tapcmd "foocmd"
Where foocmd is just our sxmo_vibratepine binary etc.