This document covers some experiments that are up for grabs but might take serious efforts and maybe not even possible.
The wm code is thankfully mostly written in a way that it can run multiple sets of rows indepedently. This is not being leveraged to full effect and a way of having multiple workspaces and group/move them would be useful.
This could be useful to map to multiple spaces as well of course, along with controls for importing / exporting between spaces.
This is a tool with a lot of potential utility. Create a zoomed out view as a minimap for quick navigation without modifying zoom would be helpful. This should also be exposable as an external connection point in order to use other devices like a streamdeck or tablet.
This should be quick and easy to implement, perhaps more difficult to find a UI language for (popup dialog?), but being able to save the current position / zoom levels in order to have a set of favorites and jump between them (with preview, but that is a more difficult thing).
To allow for mouse/touch/pen only navigation, a tool for using the builtin/ osdkbd.lua helper script would be useful. The complex part is probably to figure out which 'extra' buttons should be added for context/controls, and how to attach to expression-cell completion helpers. It should also be treated as its own overlay that is not anchored to the wm.
There is currently no persistance at all, it should be possible to save the current layout and reload it. This should re-use the scripts/ facility that testing and autorun can do.
This is a big one. It should 'for most' cases be possible to translate the rows/cells into a shell script. While there are things that will certainly be off-limits and heuristics needed to control - some pipelines would surely be possible.
It's easy to get confused by which cells are dependent on what content, Rows are supposed to help that somewhat but when other expressions are set it isn't as easy. Even for rows it should be visually indicated where there is a pipe and not.
The table that each cell expression function uses to register and integrate prepared for type- helpers, but no such helpers have been written yet. The idea was that constrained arguments, say cell addresses and so on, could have extended visual helpers that popup and expand into function arguments.
Currently a single large surface is only considered - while there are multiple paths for how hotplugging displays could work, but the more interesting is probably to combine link-target from I2 and run that through a separate rtgt that is mapped to the new displays.
Others is to mimic the durden tactic and combine I1 with some way of moving input focus between displays and have them be isolated.
Normal components, e.g. statusbar, lockscreen, ... could be added as a plugin tool that exposes a set of connection pointers per screen (bonus points if made generic enough to work as a builtin/ to upstream arcan).
Add a LANGID as a reference to possible LUTs and lookup when picking labels in popup menu or intercept set_error. Should handle more OSD Keyboard layouts as well. Arcan clients that doesn't respect GEOHINT should be updated (terminal user messages, encode- OCR and encode- T2S), and the wayland-bridge needs an xkb- table swap message.
This is in the mind-map post-it category. Having persistent notes that set visibility and scale relative to the current global anchor zoom would allow 'traditional' ZUI note taking style data organization to work. Have that handle both normal 'icons', 'post-its' and some visual partition (would probably make sense). Bonus for working at < 1fps style eInk.
A good project that spans both external processing creation and custom cell development would be something that takes a camera feed, blur/sobel and runs it through some off-the-shelf face, defines reference and exposes deviation from this reference as a trigger.