Just checking mouse.isDown works if the editor is the entirety of the
app, as is true in this fork. However, we often want to introduce other
widgets. We'd like tapping on them to not cause the selection to flash:
https://news.ycombinator.com/context?id=38404923&submission=38397715
The right architecture to enforce this is: have each layer of the UI
maintain its own state machine between mouse_press and mouse_release
events. And only check the state machine in the next level down rather
than lower layers or the bottommost layer of raw LÖVE.
I'm not sure this can trigger everywhere (I've only been able to
exercise it in Lua Carousel), but it seems like a safety net worth
having against future modifications by anybody.
This commit doesn't guarantee we'll always catch it. But if this
invariant is violated, things can get quite difficult to debug. I found
in the Lua Carousel fork that all the xpcalls I keep around were
actively hindering my ability to notice this invariant being violated.
This came up when trying to integrate my apps with the vudu debugger
(https://github.com/deltadaedalus/vudu). In general, it's a subtle part
of LÖVE's semantics that you can modify event handlers any time and your
modifications will get picked up. Now my Freewheeling Apps will follow
this norm as well.
Each one should provide a message that will show up within LÖVE. Stop
relying on nearby prints to the terminal.
I also found some unnecessary ones.
There is some potential here for performance regressions: the format()
calls will trigger whether or not the assertion fails, and cause
allocations. So far Lua's GC seems good enough to manage the load even
with Moby Dick, even in some situations that caused issues in the past
like undo.
In particular, I want to be able to switch to 'error' mode rather than
throw a real error() on test failures, because that's a little more
responsive and might be recoverable. (On some Android devices the font
is slightly different, and tests fail as a result.)
before:
stack traceback:
[string "text.lua"]:9: in function 'draw'
[string "edit.lua"]:200: in function 'draw'
[string "run.lua"]:140: in function 'draw'
[string "main.lua"]:162: in function <[string "main.lua"]:155>
[C]: in function 'xpcall'
[string "app.lua"]:38: in function <[string "app.lua"]:20>
[C]: in function 'xpcall'
[love "boot.lua"]:370: in function <[love "boot.lua"]:337>
after:
stack traceback:
text.lua:9: in function 'draw'
edit.lua:200: in function 'draw'
run.lua:140: in function 'draw'
main.lua:162: in function <[string "main.lua"]:155>
[C]: in function 'xpcall'
app.lua:38: in function <[string "app.lua"]:20>
[C]: in function 'xpcall'
[love "boot.lua"]:370: in function <[love "boot.lua"]:337>
This is all quite hacky. Many of my tests are unfortunately brittle to
changes in text rendering. Fortunately there's only one test that
currently requires a hacky special case (and a second test I tweaked
slightly to be more robust).
I can't think of a better approach. It doesn't help to standardize the
font, because version changes still come with changes to text-shaping
algorithms even if the font itself is unchanged. I could base all my
assertions on the widths of individual characters, but that would make
the tests much less readable and not express intent as clearly. So here
we are, with hopefully just a few hacky special cases (there might be a
few more as LÖVE v12 advances towards publication, and in further
versions).
To do this I need some support for multiple versions. And I need an
'error' mode to go with existing 'run' and 'source' modes
(`Current_app`). Most errors will automatically transition to 'source'
editor mode, but some errors aren't really actionable in the editor. For
those we'll use 'error' mode.
The app doesn't yet work with LÖVE v12. There are some unit tests failing
because of differences in font rendering.
Matt Wynne pointed out that snap.love would crash when a node went off
screen. While debugging it I noticed that selection1 was being set when
it shouldn't be.
Turns out I introduced a bug when I fixed the inscript bug back in June
(commit 9656e13774). One invariant I want to preserve is: selection1
should be unset after a mouse click (press and release without
intervening drag). This invariant was violated in my bugfix back in
June. I was concerned only with selection back then, and I didn't
realize I was breaking the mouse click case (in a fairly subtle way; you
can have selection set, and when it's set identically to the cursor
everything looks the same).
I think there might still be an issue in snap.love after this fix. I
noticed screen_bottom1.pos was nil, and as far as I recall that should
never happen.
This is a backport of a bugfix in pensieve.love. It's not _technically_
a bug here in lines.love, but it seems worth establishing an
architectural invariant (or rather lack of invariant).
LÖVE's standard event loop performs the following sequence of operations
in a single frame:
* process events
* update
* draw
Ideally any mutations to global state happen during the first two
phases, while drawing includes no mutation.
However, there is a special case: `starty`, the top y coordinate for
each each line in the editor. This is used all over the place, and the
cheapest way to compute it is to simply save it while drawing.
However, draw by definition only updates `starty` for lines that are
drawn on screen. To avoid stale data on lines off screen, say after
scrolling, events often clear `starty` for all lines, leaving it to the
next draw phase to repopulate the right lines.
Sandwiched between the above two "however"s, the update phase needs to
gracefully handle `starty` being nil in the occasional frame right after
an event.
I think I've audited all our uses of `starty`, and this commit fixes the
only place that violates this rule.
All the Text functions assume the cursor is always on a text line. I was
violating that invariant.
* When scrolling up, I start the cursor at the top-most line below the
screen top.
* When scrolling down, I start the cursor at the top-most line below the
screen bottom.
I think it would feel slightly more natural for it to be the
bottom-most line above the screen bottom.
However, the Text functions maintain an invariant that the bottom-most
line in a buffer will be text. There's no such invariant for the
top-most line.
Earlier the ghost while drawing wouldn't quite match the final shape.
Now the math is identical in draw_pending_shape.
It's a little unfortunate that we have this duplication of formulae.
At least there are no other stray calls of App.mouse_x in
draw_pending_shape.
Kartik K. Agaram