tynet-lichat/shirakumo

12:00:48 Colleen Shinmera: noted

12:00:48 scymtym Shinmera: noted

12:01:07 shinmera I'll probably have to buy a second hand M1 mac or something to test on that platform again, ugh

12:02:53 shinmera looks like they still go for 200-300 CHF on auction

12:03:02 shinmera for a 6 year old model, christ

12:10:32 Hugo Hello there

12:10:55 Hugo What are you working on today shinmera: ?

12:11:30 shinmera noffin

12:11:34 shinmera was busy shooting

12:13:47 shinmera now looking at bow stabilizers

12:25:15 Hugo oh right you're into archery

12:25:30 Hugo stealth archer ftw

12:33:03 shinmera https://mastodon.tymoon.eu/@shinmera/116748189886754998

12:52:58 Hugo dang remind me not to piss you off

12:53:14 shinmera don't worry, you got plenty of time to just move out of the way

13:01:48 Hugo thank God for strafing

13:02:28 Hugo seriously that's really impressive shooting though

13:09:36 shinmera thanks!

13:29:09 shinmera scymtym: pushed a fix for the animation stuff.

14:13:13 Colleen Shinmera: great

14:13:13 scymtym Shinmera: great

14:42:52 Colleen Shinmera: we have coordinate regarding input actions/events and text input component states a bit. at the moment, the action-based behavior does not consider the component state (e.g. insert vs. replace vs. isearch). on the other hand, the isearch state handles low-level events instead o factions. controller inputs are also rather low-level

14:42:52 scymtym Shinmera: we have coordinate regarding input actions/events and text input component states a bit. at the moment, the action-based behavior does not consider the component state (e.g. insert vs. replace vs. isearch). on the other hand, the isearch state handles low-level events instead o factions. controller inputs are also rather low-level

14:43:10 shinmera right

14:43:39 shinmera I just made the minimal thing to make the actions work at all-- feel free to adjust the methods again to make sense

14:44:02 shinmera though I will say, for specialisation reasons, the state argument should be first

14:44:14 shinmera rather than third

14:44:20 shinmera then overriding behaviour based on state would be much easier

14:45:05 shinmera alternatively we could just dispatch to state, if it has access to the component?

14:45:15 shinmera as in, handle event component -> handle event state

14:46:38 shinmera as for gamepad, the actions can also be translated from those, I just haven't made it convenient yet to define such mappings

14:55:22 shinmera Gonna take a look at making the clip view top align rather than bottom align

14:55:39 shinmera (wish me luck lol)

15:54:31 Colleen reordering is fine. being able to specialize to both the component and the state seems useful, though. a component subclass with specialized reactions to events would have to subclass one or more state classes and inject those subclasses into the places that make state instances otherwise

15:54:31 scymtym reordering is fine. being able to specialize to both the component and the state seems useful, though. a component subclass with specialized reactions to events would have to subclass one or more state classes and inject those subclasses into the places that make state instances otherwise

15:57:41 shinmera true

15:58:05 shinmera as for the align thing... it looks like the clip view scales the code editor to be the full size of the clip view if it would be smaller, and then the text still aligns to the bottom

15:58:09 shinmera so I can't fix that in the clip view

16:01:18 shinmera though I guess maybe I should "rip the bandaid" and just make the coordinate system grow down after all

16:01:26 shinmera would definitely break a bunchhhh of stuff

16:09:55 Colleen i still have to add vertical alignment. maybe having that would help?

16:09:55 scymtym i still have to add vertical alignment. maybe having that would help?

16:10:09 shinmera probably :)

16:14:06 shinmera the clip-view already seems to do the top alignment I want lol

16:14:11 shinmera sigh

16:14:47 shinmera scymtym: fun anecdote, in my diary two years and one day ago I wrote: "Alloy is such a mess"

16:21:57 shinmera and uh, I just noticed a new bug

16:22:32 shinmera scymtym: https://filebox.tymoon.eu/4EL7J0PFU

16:22:49 shinmera looks like clipping broke, only the "initial region" is visible, if I scroll up the rest gets clipped

16:23:21 shinmera added a clip-view example in examples/clipping.lisp for that

16:27:33 Colleen did we change anything besides ensure-visible for clip-view?

16:27:33 scymtym did we change anything besides ensure-visible for clip-view?

16:27:43 shinmera idk, I have no clue what all changed

16:28:38 shinmera the bug is not present on master, so

16:29:26 shinmera I'm going to guess that whatever culling method you implemented for text is not processing standard labels quite properly yet

16:30:57 Colleen right, the new example uses labels so the problems could be the text stuff

16:30:57 scymtym right, the new example uses labels so the problems could be the text stuff

16:40:21 Colleen the text rendering is now per-line. maybe i'm doing the renderer operations incorrectly so that things get clipped. the prepared/cache information looks identical between visible and invisible labels

16:40:21 scymtym the text rendering is now per-line. maybe i'm doing the renderer operations incorrectly so that things get clipped. the prepared/cache information looks identical between visible and invisible labels

16:42:11 Colleen removing (constrain-visibility layout renderer) in the render method of clip-view "fixes" the issue. so something about the coordinates of the visible region and the rendered shapes?

16:42:11 scymtym removing (constrain-visibility layout renderer) in the render method of clip-view "fixes" the issue. so something about the coordinates of the visible region and the rendered shapes?

16:43:39 shinmera removing constrain-visibility means no clipping is done at all, so

16:44:13 shinmera made some other fixes on the master branch related to icons and so on. Can you rebase and push?

16:44:29 shinmera I want to use that to check if it's specific to text rendering behaviour or clipping behaviour

16:46:52 shinmera ugh, sorry, sec

16:47:29 shinmera accidentally left a print in there, amended the commit

16:51:00 Colleen the new clipping example still has the same problem after rebasing

16:51:00 scymtym the new clipping example still has the same problem after rebasing

16:51:38 shinmera yea I haven't changed anything with that, I'm still putting together test cases

16:52:53 shinmera can you push the rebase?

16:54:30 Colleen i would have to force-push, right?

16:54:30 scymtym i would have to force-push, right?

16:54:56 shinmera yea it's fine

16:55:06 shinmera force-with-lease is usually the thing to do (default on magit with -f)

16:55:19 shinmera it'll prevent you from pushing over stuff you haven't fetched yet

16:55:54 Colleen sure, that's what magit uses by default. i just wanted to ask about using force at all

16:55:54 scymtym sure, that's what magit uses by default. i just wanted to ask about using force at all

16:55:56 Colleen the issue is map-visible-lines in msdf.lisp. my guess is that it does not take into account the translation from the component to its parent component

16:55:56 scymtym the issue is map-visible-lines in msdf.lisp. my guess is that it does not take into account the translation from the component to its parent component

16:56:06 shinmera yea

16:56:19 shinmera I force all the time, to amend things and rebase

16:58:18 Colleen so the clipping is fine and the general culling is fine but the text shape thinks that its lines (single line in the example) are not in the visible region

16:58:18 scymtym so the clipping is fine and the general culling is fine but the text shape thinks that its lines (single line in the example) are not in the visible region

16:58:41 shinmera right, that was my initial guess

16:59:11 shinmera also can fully confirm, displaying icons works just fine

16:59:17 shinmera even on the code-editor branch

17:02:02 shinmera okey, pushed

17:02:11 Colleen i'm pretty much fully convinced. i saw the render method getting invoked for all text shapes but only the one at 51 and below mapped over any visible lines. i suspect that the offset between the label component and layout is not considered somehow

17:02:11 scymtym i'm pretty much fully convinced. i saw the render method getting invoked for all text shapes but only the one at 51 and below mapped over any visible lines. i suspect that the offset between the label component and layout is not considered somehow

17:02:11 shinmera the clipping examples are loaded by default now

17:02:23 shinmera and the simpler example I made now interleaves labels and icons

17:04:05 Colleen should i make my own test pattern in ~/av.png?

17:04:05 scymtym should i make my own test pattern in ~/av.png?

17:04:08 shinmera scymtym: probably using the local position directly rather than using global-bounds to check for clipping

17:04:13 shinmera scymtym: oh, sorry, let me fix that

17:05:28 shinmera pushed

17:16:05 Colleen thanks

17:16:05 scymtym thanks

17:16:12 Colleen i will be busy with something else for a bit

17:16:12 scymtym i will be busy with something else for a bit

17:19:22 shinmera sure