2023-09-14 00:00:13 <thrig       > 0 or -1 traditionally
2023-09-14 00:06:17 <KipIngram   > I mean, lots of my professors stressed the importance of checking unis, using units working out as a way of having confidence in your answer, and that kind of thing.  Before finding that paper I'd have told you that's what dimensional analysis *was*.  But it turns out that hardly scratches the surface.
2023-09-14 00:06:52 <KipIngram   > You can simplify the numerical complexity of a problem (for simulation purposes and what not) by systematically reformulating in terms of dimensionless parameters, etc.  I was just blown away.
2023-09-14 00:08:18 <KipIngram   > xelxebar: You ever hear of bond graphs?
2023-09-14 00:42:55 <xelxebar    > Never heard of them! Related at all to dimensional analysis?
2023-09-14 00:58:52 <thrig       > super dimensional analysis, meanwhile, looks into the properties of Macross fortresses
2023-09-14 01:46:22 <KipIngram   > Not specifically, but... a little.  bond graphs are "domain independent" method for expressing the dynamical equations of a system.  They're particularly good when you have pieces in a system from different areas of technology, like electrical circuit mixed with pneumatics mixed with mechanical stuff etc.
2023-09-14 01:47:02 <KipIngram   > The idea is that you don't have voltage and current, force and motion, etc.  You have "through variables" and "across variables."  Through*across is always power, regardless of the underlying tech.
2023-09-14 01:47:37 <KipIngram   > It's really just the usual business of writing down differential equations from a model of a system, just with care given to not 'focus in' on a particular type of system.
2023-09-14 01:48:02 <KipIngram   > The dimensional analysis discussion made me think of it, since it's a sort of "specific units independent" thing.
2023-09-14 01:48:58 <KipIngram   > Voltage is an across variable, so is pressure.  Current is a through variable, so is flow rate.
2023-09-14 01:49:42 <KipIngram   > I've never actually done much with it, but I learned about it from a colleague in the UT Austin research lab I worked in in the early 1990's.
2023-09-14 01:50:08 <KipIngram   > I just always thought it seemed like a cool idea.
2023-09-14 02:30:42 <xelxebar    > Wait. You're located in Texas? I grew up in Corpus Christi.
2023-09-14 02:51:54 <KipIngram   > Oh, cool.
2023-09-14 02:52:24 <KipIngram   > Yes, I live a bit west of the metro Houston area.  Technically in a little town called Fulshear, but the other nearby small town Katy is a lot more well known.
2023-09-14 02:52:40 <KipIngram   > I lived in Austin from 1981 through 1996.
2023-09-14 02:53:00 <KipIngram   > Went there in 1981 to go to school.
2023-09-14 02:53:06 <KipIngram   > I grew up in Alabama.
2023-09-14 02:57:41 <KipIngram   > It's been several years since I've been down to Corpus.
2023-09-14 02:58:01 <KipIngram   > Long enough drive to "notice," you know?
2023-09-14 02:59:15 <KipIngram   > Ugh.  You know, I actually do see Norman Wildberger's points re: his concerns about irrationals.  But some aspects of it seems overly pedantic to me.
2023-09-14 03:00:02 <KipIngram   > He takes hte position that not just ANY quadratic ax^2 + bx + c = 0 can be factored into (x-r1)(x-r2)=0 form, because r1 and r2 don't necessarily exist (i.e., if they're irrational).
2023-09-14 03:00:17 <KipIngram   > But... I can get as close them as I want to, without limit.
2023-09-14 03:01:03 <KipIngram   > I think he should just propose that we replace ax^2+bx+c=0 with |ax^2+bx+c|<epsilon, with epsilon specified to suit the application at hand.
2023-09-14 03:01:16 <KipIngram   > Because you can ALWAYS find rational solutions that satisfy that.
2023-09-14 03:01:47 <KipIngram   > That makes it a completely rigorous problem AND still leaves you able to find solutions to any equation.
2023-09-14 03:04:12 <KipIngram   > I mean, that's basically how we approach it anyway.  Because if r1 and r2 are irratoinal then you never write them down and you never store them in your computer's RAM.
2023-09-14 03:04:25 <KipIngram   > We're "effectively" solving the epsilon problem as it is.
2023-09-14 03:04:51 <KipIngram   > So this would just be saying "state the problem the way we actually can solve it."  Instead of leaving to just kind of hum and look the other way.
2023-09-14 03:05:02 <KipIngram   > leaving "us"
2023-09-14 03:13:17 <MrMobius    > is there a common forth word to select from two values?
2023-09-14 03:13:40 <MrMobius    > : FOO IF SWAP THEN DROP ;
2023-09-14 03:25:15 <crc         > MrMobius: in 2016, Zarutian suggested a :? word:  : :? ( conseq altern bool ) NOT IF SWAP THEN DROP ;
2023-09-14 03:26:15 <crc         > that's the only thing I see in the logs matching this
2023-09-14 03:29:15 <KipIngram   > MAX and MIN usually are around as special cases of that.
2023-09-14 03:30:59 <KipIngram   > That seems like something easy to do if you code a word that "contains" your criterion.  Somehow passing a generic criterion in to a general purpose word seems harder.
2023-09-14 03:36:03 <KipIngram   > Oh, I switched my phone's calculator ap from WP34S to Plus42.
2023-09-14 03:36:44 <KipIngram   > https://www.thomasokken.com/plus42/images/plus42-iphone.png
2023-09-14 04:17:53 <xelxebar    > KipIngram: Dang. My parents now live in Huntsville, which sounds like it mike be close to you.
2023-09-14 04:30:22 <KipIngram   > Yes, Huntsville is an hour or so north on I45.  One of my daughters went to college there.
2023-09-14 04:30:59 <KipIngram   > She's in grad school in North Carolina now.
2023-09-14 05:16:31 <xelxebar    > That's nuts. My mom worked for Sam Houston for 10+ years in admissions. Not unlikely that your daughter spoke with her.
2023-09-14 05:19:03 <thrig       > small something something world something after all
2023-09-14 06:02:03 <KipIngram   > Wow - that really is a small world.
2023-09-14 10:19:10 <veltas      > Huh cannot ssh from work anymore
2023-09-14 10:19:22 <veltas      > They're onto me lol
2023-09-14 10:27:05 <veltas      > Oh no looks like my VM just lost internet/interest
2023-09-14 10:27:14 <veltas      > (work VM, not oracle)
2023-09-14 11:49:14 <hello-operator> MrMobius: mux is one such word, it is a bit more generic
2023-09-14 11:49:54 <hello-operator> : mux dup >r and swap r> invert and or ; ( x1 x2 mask -- x )
2023-09-14 11:50:45 <hello-operator> Short for "multiplex", based off of the operator
2023-09-14 11:50:53 <hello-operator> You can define min/max with them
2023-09-14 11:51:27 <hello-operator> 2dup <TEST> mux
2023-09-14 12:21:04 <veltas      > I think someone called is CHOOSE
2023-09-14 12:21:07 <veltas      > it*
2023-09-14 12:21:17 <veltas      > I think this same question gets asked at least once a year as well
2023-09-14 12:41:02 <rendar      > dup is ( x1 -- x1 x1 ) and 2dup is ( x1 x2 -- x1 x2 x1 x2 ), in my forth i'm adding dup2 which is ( x1 -- x1 x1 x1 ) and dup3 which is ( x1 -- x1 x1 x1 x1 ) ...with this notation also 2dup2 would be contemplated, thoughts?
2023-09-14 13:09:26 <xelxebar    > Depending on how true/false are represented, PICK might get you part way to a choose.
2023-09-14 13:10:30 <xelxebar    > : CHOOSE ( x y cond -- x|y ) PICK NIP NIP ;
2023-09-14 13:18:19 <veltas      > CHOOSE or CHOICE
2023-09-14 13:33:01 <xelxebar    > Tangential, but IMHO, some traditional Forth naming choices are kind of clunky. Like `... STATE IF ... THEN` doesn't read particularly well.
2023-09-14 13:53:21 <gordonjcp   > xelxebar: well, you can change that
2023-09-14 13:53:29 <gordonjcp   > xelxebar: how would you like it to read?
2023-09-14 13:54:24 <veltas      > You mean Forth Inc, because fig Forths used ENDIF
2023-09-14 13:54:37 <veltas      > Precisely because they thought it read weird
2023-09-14 13:54:45 <gordonjcp   > KipIngram: re solving for <epsilon, it's one of the most important concepts that people getting into electronics don't get taught
2023-09-14 13:54:57 <gordonjcp   > it's vitally important to know how to work out things like Ohm's Law
2023-09-14 13:55:02 <veltas      > I personally think it makes sense as long as you understand it's "THEN" in a different semantic meaning to "THEN" in most if statements
2023-09-14 13:55:05 <gordonjcp   > it's even more important to know what you can get away with
2023-09-14 13:56:58 <xelxebar    > veltas: Well, I was mostly thinking STATE IF, more than THEN. Wouldn't a dumb rename to COMPILING? or whatnot would make sense?
2023-09-14 14:18:04 <veltas      > xelxebar: Or IMM or something, I suppose you're right that STATE's quite generic
2023-09-14 14:18:21 <veltas      > But you can always hide STATE if you want to use the name for something else
2023-09-14 16:58:12 <KipIngram   > xelxebar: COMPILING would make more sense given what STATE actually is used for.
2023-09-14 16:59:25 <KipIngram   > But STATE has been in Forth for so long that everyone is just "used to it" - it's absorbed its meaning pretty thoroughly.
2023-09-14 20:47:37 <KipIngram   > Oh, earlier I suggested s[i] as a way for an APL expression to access the Forth stack.  Along with that, I might use f[i] to access specific slots in an open stack frame.  Two slightly different but quite similar ideas.
2023-09-14 21:59:12 <rendar      > dup is ( x1 -- x1 x1 ) and 2dup is ( x1 x2 -- x1 x2 x1 x2 ), in my forth i'm adding dup2 which is ( x1 -- x1 x1 x1 ) and dup3 which is ( x1 -- x1 x1 x1 x1 ) ...with this notation i was thinking 2dup2 possible, do you think its clear?
2023-09-14 22:01:42 <thrig       > is that something you do enough to need a core word for?
2023-09-14 22:03:51 <rendar      > well, yes
2023-09-14 22:03:57 <rendar      > otherwise i didn't
2023-09-14 22:04:14 <rendar      > i have functions that pushes 2 things into the stack, so drop2 is very usefsul instead of drop drop
2023-09-14 22:05:09 <rendar      > sorry, i meant dup, not drop
2023-09-14 22:21:31 <KipIngram   > I think it's very much an individual call whether to have double precision int words.
2023-09-14 22:21:53 <KipIngram   > Made a lot of sense in 16-bit systems, and a little even in 32-bit systems.  I've never seen the need in 64-bit systems.
2023-09-14 22:22:00 <KipIngram   > But... ymmv.
2023-09-14 23:06:06 <xelxebar    > KipIngram: Good point about the vectors like 1 2 3. Are you familiar with strand notation? It's a kind of generalization. Back-to-back expressions create an vector with those elements, e.g. (1 2 3) (4 5 6) or 'APL' 'in' 'Forth' or even (1 2 3) 'APL'.
2023-09-14 23:07:43 <xelxebar    > J doesn't have strand notation and recognizes strings of space-separated numbers as a single *token*.
2023-09-14 23:08:43 <xelxebar    > BQN (https://mlochbaum.github.io/BQN/) decides to make space always mean "function application" and offers a couple different decorations for array notation.
2023-09-14 23:10:12 <xelxebar    > https://mlochbaum.github.io/BQN/doc/arrayrepr.html#array-literals
2023-09-14 23:11:04 <xelxebar    > That's probably a relatively tractable problem.
2023-09-14 23:11:23 <xelxebar    > The bigger issue, as you have mentioned a couple times, seems to be infix notation.
2023-09-14 23:12:54 <xelxebar    > Removing infix like you mention is way too drastic, I think. It would really disrupt the way APL expressions read.
2023-09-14 23:14:14 <xelxebar    > What about going the other way? Forth syntax is so regular that it's probably easier to have a "Forth mode" inside APL.
2023-09-14 23:15:32 <xelxebar    > If one could stomache Polish Notation instead of RPN, then a "put array elements on stack" operation, say ARR->STK, might work
2023-09-14 23:17:05 <xelxebar    > Visually, it'd be something like this: ... R@ >R DUP ARR->STK (+/÷≢) ⍳100
2023-09-14 23:18:17 <xelxebar    > The RPN to PN switch is in Forth is equally disorienting, though. It's like we're bumping up against the problem of mixed RTL and LTR orthographies.