2023-10-16 01:44:14 <hello-operator> MrMobius: Cheers! I'll take a look!
2023-10-16 02:19:49 <KipIngram   > I'm watching a suite of videos on electric motor design.  Brings back some old memories.  Great application for finite elements.
2023-10-16 02:20:19 <KipIngram   > Specifically this set is on switched reluctance motor design, but it looks like they've got video sets on a bunch of different kinds.
2023-10-16 02:21:57 <KipIngram   > I've always found finite element methods really cool.
2023-10-16 02:27:20 <KipIngram   > Pretty simple kind of motor - basically looks like this inside:
2023-10-16 02:27:22 <KipIngram   > https://www.eeeguide.com/wp-content/uploads/2019/01/Switched-Reluctance-Motor.jpg
2023-10-16 02:28:10 <KipIngram   > The rotor pole is pulled toward alignment with whichever stator pole is energized at the time, and the pole count on the stator and rotor is such that there's always a clear "next pole" to turn on to continue the motion.
2023-10-16 02:28:28 <KipIngram   > So the controller just activates the windings in sequence, and the rotor spins up.
2023-10-16 02:29:37 <KipIngram   > In some ways all of this has gotten a lot simpler since we got microcontrollers - before that you had to work out some sort of angle dependent switching mechanism.  But with a uC you can just do pretty much anything.
2023-10-16 02:30:51 <KipIngram   > Switched reluctance is one of the easier to understand motor types.
2023-10-16 05:30:33 <KipIngram   > Ah, this motor design involves a semi-interesting optimization method.  The software lets you enter all the relationships among an arbitrarily chosen set of variables, depending on whatever it is you're doing.  But you can also flag variables as "optimizable" or not (some of them will just depend on variables you are optimizing over, and you don't want to independently change those).  Then when you have it
2023-10-16 05:30:35 <KipIngram   > all set up the software has the info it needs to run some general optimization procedure.
2023-10-16 05:30:47 <KipIngram   > Guess it's really more about how the interface works than how the calculations are done.
2023-10-16 05:30:59 <KipIngram   > Those extra layers of variables sort of become 'convenience variables.'
2023-10-16 05:31:14 <KipIngram   > But don't add any  new degrees of freedom.
2023-10-16 05:31:46 <KipIngram   > In a Forth system each of those could have a corresponding word that calculates them.
2023-10-16 05:32:21 <KipIngram   > You could collect them all into a self-contained vocabulary, and the optimization procedure could run against that vocabulary  name.
2023-10-16 05:32:37 <KipIngram   > It could update all of the variable values in the vocabulary.
2023-10-16 05:32:46 <KipIngram   > It's a nice way of "containing" a piece of work.
2023-10-16 05:33:14 <KipIngram   > I like ideas that leverage the dictionary.
2023-10-16 05:34:29 <KipIngram   > You could also put the optimizable variables in a particular vocabulary within the "problem vocabulary," rather than setting per-word flags.  That would segregate the variables just as well as flags.
2023-10-16 05:36:13 <KipIngram   > I'm watching this video and looking at all the morass of GUI widgets that are involved in telling the software how to know what to do - neat that a plain old Forth vocabulary structure could do it just as well.
2023-10-16 05:51:36 <KipIngram   > After all, a vocabulary is a linked list, and in each case the optimization variable itself would be at address *(*(p+cell)).  Probably about as efficient as any method of identifying a particular subset of a collection of variables.
2023-10-16 05:52:29 <KipIngram   > Oh, wait - that would be the value of the variable.  Drop the leading * to get the address.  And of course that's dependent on my implementation details.
2023-10-16 05:53:48 <KipIngram   > Anyway, the new wrinkle here that I hadn't thought of before is to actually use the dictionary organization as pertinent to the actual algorithm you're running, rather than just as a way to find words at compile time.
2023-10-16 17:08:12 <rendar      > has some old Forth a kind of macro concepts, like in Lisp?
2023-10-16 17:11:42 <GeDaMo      > https://www.atariwiki.org/wiki/Wiki.jsp?page=Forth%20Macros
2023-10-16 18:04:22 <KipIngram   > rendar: In Forth's very early days its "definitions" worked more like macros - the system would just drill through the words on an interpreted line, and go interpret the contents of each.  That was recursive until it got down to primitives.
2023-10-16 18:04:39 <KipIngram   > In other words, nothing was actually "compiled" in those days - it was all kind of text-based.
2023-10-16 18:04:47 <KipIngram   > But that's really old stuff.
2023-10-16 18:04:52 <rendar      > yeah indeed
2023-10-16 18:05:22 <KipIngram   > I'm not familiar with a Forth that has what you'd call macros these days - compiled definitions do more or less the same job and do it with much higher performance.
2023-10-16 18:05:32 <rendar      > my Forth works basically only with strings, and some kind of bytecode to identify native builtin words
2023-10-16 18:05:56 <KipIngram   > But of everyone around here I may have the least familiarity with the "existing Forth flora and fauna."
2023-10-16 18:07:03 <KipIngram   > You can see some of this here - it's a self-published book Chuck wrote way back in the primitive days:
2023-10-16 18:07:05 <KipIngram   > http://forth.org/POL.pdf
2023-10-16 18:07:15 <rendar      > cool thanks
2023-10-16 18:08:21 <KipIngram   > Chuck was also a bit more of an "apologist" then than later on.  For example, he goes to some length in there to show how Forth can be used to parse infix arithmetic expressions.  He seemed to feel that "lack of infix" would be thought of as a weakness, and wanted to show how to get around it.
2023-10-16 18:08:32 <KipIngram   > Later on his attitude was more "you shouldn't want to do that."
2023-10-16 18:08:48 <KipIngram   > Guess we all get more obstinate as we age.
2023-10-16 18:08:56 <KipIngram   > Or more confident, or both.
2023-10-16 18:42:28 <rendar      > i guess that the postfix notation is great for parsing
2023-10-16 18:42:34 <rendar      > basically you don't have any parsing
2023-10-16 18:42:44 <rendar      > we can say that Forth is the only language without any syntax
2023-10-16 19:39:12 <KipIngram   > Yes, Forth "solves the problem" by mostly making the problem go away.
2023-10-16 19:40:00 <KipIngram   > And yeah, the "no syntax" claim is 99% true.  You do have things like IF ... THEN and BEGIN ... AGAIN, and I guess those qualify as "syntax."  But those really are corner cases - by and large it's syntax free.
2023-10-16 19:40:32 <KipIngram   > definition = : <name>  <words> ;
2023-10-16 19:40:37 <KipIngram   > That's also a touch of syntax.
2023-10-16 19:41:51 <KipIngram   > Gotta love this:
2023-10-16 19:41:53 <KipIngram   > https://www.nature.com/articles/d41586-023-03238-5
2023-10-16 19:42:23 <KipIngram   > People are using ChatGPT to write grant applications - one example is cited of cutting three days of writing work down to three hours.
2023-10-16 19:42:40 <KipIngram   > Opinions seem to vary from "that's horrible" to "so what?"
2023-10-16 19:43:03 <KipIngram   > I think the biggest problem with it is that you can't even be sure your grant application author even  knows what's in his application.
2023-10-16 19:43:35 <KipIngram   > This is just one more example of Pandora's box being open; we're never go in to be able to close it again now.
2023-10-16 19:43:58 <KipIngram   > Just think about how much English essay homework is no doubt being done with ChatGPT.
2023-10-16 19:44:41 <KipIngram   > You'd think, though, that defending against that should be easy - a teacher ought to be able to just shove some of a kid's paper into ChatGPT and have it TELL him or her whether ChatGPT wrote it or not.
2023-10-16 19:44:49 <KipIngram   > I'm sure it has a record of what it's written.
2023-10-16 19:57:18 <Zarutian_iPad> KipIngram: what was that thing, ah, yes Grammarly, that shows up to.
2023-10-16 20:09:05 <KipIngram   > Yes, I remember that I think.
2023-10-16 20:11:29 <KipIngram   > rendar: In that Chuck book I linked you to earlier, I do think the actual algorithm he describes for parsing infix is interesting in its own right - the technique probably has applications in other areas.
2023-10-16 20:12:04 <KipIngram   > Scattered around in there he mentions several little ideas that we don't normally see in Forth these days but nonetheless might have some utility from time to time.
2023-10-16 20:18:04 <KipIngram   > So, I decided to upgrade my telescope controller handset rather than piddle around with RS232.
2023-10-16 21:49:03 <KipIngram   > Wow - the humidity today is just 36%.  That's VERY low for Houston.
2023-10-16 22:37:51 <rendar      > KipIngram, thanks, very interesting inputs
2023-10-16 22:38:17 <rendar      > oh are you from Houston? it should be cool to live there :)
2023-10-16 22:53:14 <KipIngram   > Yes; I don't live literally in the city limits, but it's quite close by.
2023-10-16 22:53:24 <KipIngram   > It's got a lot of things to offer.
2023-10-16 23:47:23 <immibis     > the same humidity is a lower percentage when it's hotter, and it's the hottest on record for this time of year (a statement which has been true every day since june)