[RFC] client: Rewrite the client stacking code in Lua. #3751
Conversation
Elv13
force-pushed
the
client_stacking
branch
2 times, most recently
from
8432624
to
17f062e
Compare
Elv13
force-pushed
the
client_stacking
branch
2 times, most recently
from
fac155d
to
22a3461
Compare
Codecov Report
@@ Coverage Diff @@
## master #3751 +/- ##
==========================================
+ Coverage 90.97% 91.01% +0.04%
==========================================
Files 900 900
Lines 57500 57553 +53
==========================================
+ Hits 52309 52381 +72
+ Misses 5191 5172 -19
Flags with carried forward coverage won't be shown. Click here to find out more.
|
There was a problem hiding this comment.
All this seems pretty good, and at first glance it works fine too. I'm sending this message from this branch right now, actually.
The only things that I'd point out other than the changes i have requested, is that spawning windows feels somewhat sluggish compared to the main branch, but I don't think that that's a dealbreaker for most people lol
| for _, c in ipairs(clients) do append(c) end | ||
| for i=#drawins, 1, -1 do append(drawins[i].get_wibox()) end |
There was a problem hiding this comment.
These two lines should be swapped, as in the original C code, drawins are rendered under clients. (plus it kinda ruins the point of ontop)
There was a problem hiding this comment.
Not sure about this, swapping those line breaks quite a few things. For example, the client go on top of the wibar in the "old" builds. If I swap these, the wibox goes on top. That doesn't mean it didn't break anything. It's after all not exactly what it did before. What did you witness being in the wrong order?
There was a problem hiding this comment.
That must've been a mistake on my end. Wiboxes do indeed go on top when these two lines are swapped. I just thought that it would be the other way around. Didn't notice line 482 inserting into the table at index 1 lol
There was a problem hiding this comment.
IMO client should go over wibar only if it have c.ontop == true
This adds 2 new requests:
* request::raise: Helps to decouple ewmh and client class
internally plus allow some focus stealing policies.
* reqest::restack: Send *why* something is restacked to Lua. It
allows to do things like sending a client to the back of a
layout rather than use the master area.
This is mostly a 1:1 port of the C code. The idea is to use this as
a starting point to have a stack per `tag` rather than something
global. It also paves the way for stacking wibox among clients
in a predictable way. None of that is exposed in the public API
as part of this commit. The point is to get enough going so the
wibox desktop layer PR can be implemented on top of this.
Elv13
force-pushed
the
client_stacking
branch
from
22a3461
to
c6a9148
Compare
| * \param c The client to push. | ||
| * \param context An human readable reason of why this was done. |
There was a problem hiding this comment.
| * \param context An human readable reason of why this was done. | |
| * \param context A human-readable reason of why this was done. |
Though, I'd disagree that this is "human-readable". The only values for this I've seen in the PR are cryptic, single word strings like "maprequest".
If the context is supposed to be human-readable, i.e. printed, it should at the very least include the full caller's function name, so that you have something to search for.
Or if context is supposed to be compared to programmatically, it should be an enum, with each value documented.
This adds 2 new requests:
This is mostly a 1:1 port of the C code. The idea is to use this as a starting point to have a stack per
tagrather than something global. It also paves the way for stacking wibox among clients in a predictable way. None of that is exposed in the public API as part of this commit. The point is to get enough going so the wibox desktop layer PR (#3750 ) can be implemented on top of this.