2022-05-01 13:33:35 <KipIngram   > One little glitch popped up in the EXPECT stuff last night.  Later on I noticed that the "buffer full" test wasn't working right anymore.  I stopped adding bytes to the buffer, but the screen got fouled.
2022-05-01 13:35:15 <KipIngram   > That full test was at the beginning of the word that inserts new characters.  I pushed that insert word down a level.  The test sent me back to that new word instead of where it was designed to send me.
2022-05-01 13:35:34 <KipIngram   > Anyway, I just moved the test word up into the new layer, and that fixed it.  But it did have to be "touched."  :-)
2022-05-01 15:18:54 <KipIngram   > Hey, does anyone happen to know if the ANSI sequences for cursor position save and restore can be "nested"?  Or is it just "one deep"?
2022-05-01 15:19:24 <eris[m]     > pretty sure its only one deep
2022-05-01 15:19:53 <KipIngram   > Yeah - somehow I've a nagging feeling of that as well.
2022-05-01 15:20:40 <KipIngram   > Guess it's nice it's there at all.
2022-05-01 19:08:24 <KipIngram   > Hey - here's a neat trick.  Say you have a 64-bit item with a utf-8 char in the low bytes, and zero otherwise.  You need to insert that into a fixed size buffer of text, but you don't want to unless doing so won't push any "non-space" bytes off the end of the buffer.
2022-05-01 19:08:50 <KipIngram   > Here's how to do it.  Grab the last four bytes of the buffer, and xor that with 0x20202020.
2022-05-01 19:09:10 <KipIngram   > Now form the "byte-wise inner product" of that with your insert char bytes.
2022-05-01 19:09:23 <KipIngram   > If the total inner product result is zero, you are safe to proceed.
2022-05-01 19:09:38 <KipIngram   > I think that will work regardless of how many non-zero bytes you have in your char cell.
2022-05-01 19:09:47 <KipIngram   > Well, up to four.
2022-05-01 19:10:39 <KipIngram   > If a buffer byte isn't a space, you'll get non-zero in that byte of the xor.  But if you don't need that many bytes, then that byte of your incoming will be zero.
2022-05-01 19:11:22 <KipIngram   > I like how that combines logic and arithmetic in a useful way.
2022-05-01 19:11:42 <KipIngram   > I was really chafing at the idea of basically having to do four separate tests.
2022-05-01 19:12:59 <KipIngram   > I've spent the day today tweaking READ so that it works as needed in EXPECT, but also can be called to let you edit a fixed size buffer anywhere in RAM.
2022-05-01 19:14:53 <KipIngram   > That's all done now except for implementing the test I just described.  Right now it just "cliff pushes" bytes off the high end, regardless of what they are.
2022-05-01 19:15:12 <KipIngram   > Which isn't actually a "wrong" behavior; just a little undesirable.
2022-05-01 19:15:58 <KipIngram   > At first it was pushing a byte or two into the space just beyond the buffer, but I fixed that.