Highlighter regression changes display encoding of characters
Completed
A recent release fixed the “jitter” issue that could happen when typing in certain longer structures (e.g. a long list that extends above the top of the window.)
Long story short, the “jitter” seems to be related to refreshing the text attributes of a large chunk of text, as the macOS text layout engine catches up.
The recent fix tried to limit this refresh purely to ranged of changed characters, but I just discovered another edge case where this fails, beyond those initially tested for. To recreate, use:
> plain text line
> (empty line)
> (tab)this is a code block
Place the insertion point on the empty line and hit delete. The code block line is converted to regular text, which changes the font and then displayed character encoding of “this is a code block”. The actual text is unchanged. You can verify this by typing a new character on the modified line and see that the original text is recovered.
I'll need to keep digging to try and find an acceptable middle ground between performance and accuracy…
In the meantime, if you find a different example of what appears to be the same issue, feel free to add it to this thread.
-
Sometimes writing something down gives you a new idea.
I'll push a new release after a bit more testing that fixes this particular example, and should fix the most common similar examples, but still prevents the "jitter"."
There may still be some other edge cases that it misses. Please let me know if you find one.
0 -
Closing due to lack of reported issues. Please open a new thread if something comes up.
0
Please sign in to leave a comment.
Comments
2 comments