Snap the TextBox cursor to the pixel grid #1794
Conversation
CryZe
force-pushed
the
snap-cursor-to-grid
branch
2 times, most recently
from
24a5883 to
eb9023e
Compare
This snaps the cursor of the TextBox to the grid to ensure it's always rendered as sharp as possible.
CryZe
force-pushed
the
snap-cursor-to-grid
branch
from
eb9023e to
22de5f8
Compare
xarvic
added
S-ready
|
I'd like to take a look at this before it gets merged please. (I can't figure out how to self-request a review from my phone 😕) |
xarvic
added
S-needs-review
There was a problem hiding this comment.
Thanks for this! I do think it's important to get this stuff right, and my main concern is totally unrelated to this patch, and is more about how need some mechanisms for being more consistent about this, across the board. There's a discussion in #1088, and there's been various efforts in the past (linebender/kurbo#93, linebender/kurbo#107, #904).
Anyway, main actual question is whether we should round to trunc?
| cursor.p0.x = cursor.p0.x.trunc() + 0.5; | ||
| cursor.p1.x = cursor.p0.x; |
There was a problem hiding this comment.
- this should be correct for coregraphics and skia; i'm not sure if directwrite behaves the same way?
- might
roundmake more sense thantrunc? - I'm actually also not sure if this will work at all scale factors. it should work at 1x and 2x? I'm less sure about other cases.
All that said: I'm not opposed to this at all, it's just exposing some of my general confusion about pixels and scale factor. I would really love to see a sort of 'unified' theory of this stuff; until we have that it's going to feel like our drawing code is an inconsistently applied series of heuristics. :/
|
I'm going to merge this, it's an improvement on the current code and we can always fix other problems later. |
This snaps the cursor of the TextBox to the grid to ensure it's always rendered as sharp as possible.
Screenshot:
Top is with grid snapping, bottom without.