2023-12-08 04:33:05 <MrMobius    > I think eforth works in esp32
2023-12-08 04:35:27 <MrMobius    > there's a lot of interest im svfig
2023-12-08 04:40:20 <MrMobius    > also some of the newer esp32s are not xtensa
2023-12-08 07:24:44 <crc         > I'm not really worried about the "best forth" part; my forth runs on a virtual machine, so I just need to implement that.
2023-12-08 13:04:58 <KipIngram   > So, what exactly characterizes a funnel shifter?  My wild guess is that it somehow achieves an arbitrary shift by concatenating power-of-two bit count shifts based on the bits of the count specification?
2023-12-08 13:05:34 <KipIngram   > I.e., it has the hardware connections to do any 2^n shift, but not arbitrary n shifts?
2023-12-08 13:05:55 <KipIngram   > That does seem like a reasonable middle ground between single bit shift and barrel shifting.
2023-12-08 13:08:52 <KipIngram   > Ah, I found some info on them here:
2023-12-08 13:08:53 <KipIngram   > http://people.ee.duke.edu/~jmorizio/ece261/classlectures/arithcircuits2.pdf
2023-12-08 13:09:10 <KipIngram   > And my wild guess looks totally wrong.
2023-12-08 13:12:50 <KipIngram   > So, is there perhaps a C library that gives you a routine that will take from you a string representing an x86_64 assembly language instruction, and will return to you an array of bytes for the encoding of that instruction?
2023-12-08 13:13:18 <KipIngram   > That seems like something that ought to exist.
2023-12-08 13:18:07 <thrig       > maybe something in capstone or somesuch?
2023-12-08 13:19:16 <KipIngram   > Looks like maybe there is:
2023-12-08 13:19:18 <KipIngram   > https://intelxed.github.io/ref-manual/
2023-12-08 13:20:15 <KipIngram   > Doesn't look like it works exactly as I described, but it seems to do the basic job.
2023-12-08 13:20:50 <unjust      > also, unfortunately not written in C (but if you're linking to a library who cares if it's written in C++): https://asmjit.com/
2023-12-08 13:20:53 <KipIngram   > Anyway, this C emulator/generator I've talked about - at some point I'm going to need to add the actual vm implementation to the image.
2023-12-08 13:21:04 <KipIngram   > And it seems like a tool like this could  help with that (for x86 targets).
2023-12-08 13:21:13 <thrig       > C++ can be annoying to compile and annoying to link to
2023-12-08 13:21:15 <KipIngram   > Thanks.
2023-12-08 13:22:17 <unjust      > thrigh: it's possible to make compiling and linking to code written in C terrible too
2023-12-08 13:22:22 <unjust      > thrig*
2023-12-08 13:23:39 <thrig       > sure, but it's easier to get "special" compile times and unusable symbols from C++
2023-12-08 13:24:01 <unjust      > fair enough
2023-12-08 13:26:08 <thrig       > and I've seen a bunch more "there's no compiler for that yet" due to C++23 or whatever than for folks using too-modern C
2023-12-08 16:23:09 <zelgomer    > what do you guys like for max column width when writing forth?
2023-12-08 16:24:06 <zelgomer    > i know classic worth was 64 columns. when i write c i try to target 80. thinking of splitting the difference and restricting myself to 72
2023-12-08 16:34:55 <thrig       > something less than the 80 the terminals are set to, anyways
2023-12-08 17:17:01 <Zarutian_iPad> 40 chars
2023-12-08 17:33:23 <KipIngram   > I usually limit myself to 64 chars max, but start to squirm a little if a defition get longer than 45 characters or so.
2023-12-08 17:33:49 <KipIngram   > I use the shortest names I can, though, so if you're a longer name person that would probably come in more like 64.
2023-12-08 17:34:27 <KipIngram   > I lived in the  "block = 16 64-char lines" world long enough to get it kind of burned in.
2023-12-08 17:34:27 <thrig       > orthodoxy from other programmers is to wrap at 120 or even more columns for modern languages
2023-12-08 17:34:46 <KipIngram   > I despise wrapped lines for program source.
2023-12-08 18:22:58 <zelgomer    > KipIngram: most of my definitions don't span a full 64 columns, but because many of my definitions are so short and repetative, i often like to write multiple things on one line
2023-12-08 18:23:03 <crc         > zelgomer: I've gotten into the habit of limiting lines to a max of 64 characters, including for languages other than Forth.
2023-12-08 18:23:18 <zelgomer    > to columnate lots of similar definitoins, for example
2023-12-08 18:25:28 <crc         > I occasionally have more than one word per line, if they're short enough and it doesn't make the source too dense
2023-12-08 18:26:52 <KipIngram   > zelgomer: I've done that in C sometimes.  Usually if a few very short statements are "functionally integrated" with one another.
2023-12-08 18:27:20 <KipIngram   > And once in a while I've done it in Forth too, when it made the difference whether a set of source would fit in a block.
2023-12-08 18:27:36 <KipIngram   > On my older Forth systems they were 64-char lines, period, even if i didn't use them all.
2023-12-08 18:27:45 <KipIngram   > This last go round I included linef feeds, though.
2023-12-08 18:28:52 <crc         > I mostly do this with variables, occasionally words (e.g., http://forth.works/share/ap8tXbLqcw.txt )
2023-12-08 18:29:41 <KipIngram   > Yes, I've done it with variables sometimes too.  I shorten it to var
2023-12-08 18:29:54 <KipIngram   > And voc rather than vocabulary.
2023-12-08 18:30:21 <crc         > I have var and var-n (to set an initial value)
2023-12-08 18:30:50 <KipIngram   > Ah, nice.  I might call that var!
2023-12-08 18:41:21 <zelgomer    > yeah i call that var! or variable!
2023-12-08 19:13:06 <Zarutian_iPad> why I like to use 40 chars wide is twofold: I can use the other 40 or so char for comments
2023-12-08 19:13:44 <Zarutian_iPad> be it stack diagrams or notes on why the code is as it is
2023-12-08 19:19:09 <zelgomer    > psh comments? ok boomer
2023-12-08 20:04:20 <crc         > comments are good. I don't really do inline comments other than stack effects. On my system using files, I have a system of comments, code sections, and test sections. On the block based system, I either have full line comments or keep commentary in separate blocks.