freenode/#clim - IRC Chatlog

5:32:08 beach Good morning everyone!

5:46:57 loke Hello beach!

5:49:12 beach How is the editor going?

6:43:37 loke beach: Still thinking of display optimisation

6:44:10 loke I'm coming to the conclusion that the “items” in the LINE should be grapheme clusters, not Unicode codepoints.

6:44:30 loke This is because the items describe individual drawn elements, as well as represents the way the cursor moves.

6:44:51 loke It's not ideal, but the alternative is worse.

7:03:34 beach I have to trust you on that one.

7:21:04 loke Don't. :-)

7:21:09 loke there are many ways to implement this stuff

7:21:12 loke How di dyou do it?

7:27:51 beach I didn't think much about Unicode at all. Mainly because my knowledge is very limited.

7:28:08 beach My goal was to create an excellent editor for Common Lisp source code.

7:28:34 beach So I have given much more thought to incremental parsing and analysis.

7:28:43 loke Fair enough, but even without the added issues of Unicode, you must have worked on displa optimisation? I'm pretty sure you're not repaininting the entire window whenever a button is pressed.

7:29:08 beach That is not a problem anymore.

7:29:46 loke How did you habndle it?

7:30:37 beach I didn't. And it's plenty fast.

7:31:09 loke You mean every command reslted in a full repaint of everything?

7:31:32 loke That would cause a lot of flickering, no?

7:31:35 beach What I did, though, was to explicitly manipulate the output history.

7:31:45 loke I see

7:31:54 beach Well, I count on double buffering to fix the flickering.

7:32:14 loke In my implementation I'm not using an output history

7:32:15 beach So I don't do a redisplay.

7:32:43 beach Oh?

7:33:18 beach So you must then be re-implementing the repaint protocol.

7:33:30 loke No. I directly implement HANDLE-REPLAINT

7:33:33 loke Correct

7:33:59 beach What was your reason for doing that? McCLIM does that quite well.