Restore workspaces to output when re-enabled #2115
Conversation
RedSoxFan
force-pushed
the
restore-workspaces
branch
3 times, most recently
from
c0543ca
to
7b753bb
Compare
RedSoxFan
changed the title
|
This has been rebased and is ready for review |
|
Not a fan of this approach. I'd rather have each workspace just keep a track of the outputs it has been on in a list of priority. That way, you can remove N (where N>1) monitors and plug them back in and have everything as it was. |
Sounds good to me. I'll update the PR. I'll need some help testing the implementation for that approach though since I only have two monitors. |
RedSoxFan
force-pushed
the
restore-workspaces
branch
from
7b753bb
to
d1c7403
Compare
sway/tree/container.c
Outdated
| @@ -433,6 +445,9 @@ void container_descendants(struct sway_container *root, | |||
| func(item, data); | |||
| } | |||
| container_descendants(item, type, func, data); | |||
| if (i < root->children->length && root->children->items[i] != item) { | |||
There was a problem hiding this comment.
Hmm, this is a bit hacky, and only works if the current container is removed. I'd rather copy the list when doing this sort of thing.
There was a problem hiding this comment.
I would prefer the PR be changed to use a double/nested for loop to iterate the workspaces, rather than use container_descendants. We don't need to walk the entire tree.
If we weren't going to go with that suggestion then my next preference would be to copy container_descendants into a new function called container_descendants_safe and go with the list copy idea.
There was a problem hiding this comment.
I would prefer the PR be changed to use a double/nested for loop to iterate the workspaces
Well, that doesn't change the fact that the inner loop will iterate over a list that will be mutated. But I agree it's a better idea.
There was a problem hiding this comment.
I switched it to using a nested for loop. Since it's only ever removes the current element, it should be fine.
| list_insert(ws->output_priority, old_index, strdup(output->name)); | ||
| } else if (new_index > old_index) { | ||
| char *name = ws->output_priority->items[new_index]; | ||
| list_del(ws->output_priority, new_index); |
There was a problem hiding this comment.
Do you need to free the name in the output_priority list before deleting it?
There was a problem hiding this comment.
No, because instead of using strdup(output->name) like I do if it's not in the list, I'm just re-inserting the same copy on the following line. I could do the following, but I felt it was a waste to make a duplicate after freeing a duplicate:
char *name = ws->output_priority->items[new_index];
list_del(ws->output_priority, new_index);
free(name);
list_insert(ws->output_priority, old_index, strdup(output->name));
```There was a problem hiding this comment.
Ah, yep. I had a derp moment. It was 12:30am local time.
sway/tree/container.c
Outdated
| @@ -433,6 +445,9 @@ void container_descendants(struct sway_container *root, | |||
| func(item, data); | |||
| } | |||
| container_descendants(item, type, func, data); | |||
| if (i < root->children->length && root->children->items[i] != item) { | |||
There was a problem hiding this comment.
I would prefer the PR be changed to use a double/nested for loop to iterate the workspaces, rather than use container_descendants. We don't need to walk the entire tree.
If we weren't going to go with that suggestion then my next preference would be to copy container_descendants into a new function called container_descendants_safe and go with the list copy idea.
RedSoxFan
force-pushed
the
restore-workspaces
branch
from
d1c7403
to
e2b2fb0
Compare
|
Thanks! |
This is forked from #2108 so that needs to be merged first
In #2108, it was requested that workspaces be restored to their outputs when re-enabled as discussed in swaywm/wlroots#479. This PR implements that request.