2023-03-26 13:14:35 <nature      > Hello! So I got further on my little VM project, I now have fully working screen (can draw pixels) and controller input, I just lack some audio stuff and I am wondering wether to handle args from the cmdline in a special case to make it convenient to access them.
2023-03-26 13:15:19 <nature      > Now, I would be very happy to get some feedback on my set of base words (which are like the assembly for my vm).
2023-03-26 13:16:16 <nature      > I have no experience designing opcodes for a vm, I just started with the colorforth/f18a opcodes and then added a few like "/" "mod" "swap" etc...
2023-03-26 13:17:01 <nature      > it's a stack machine, with a circular data stack of 8 slots + a tos, and the return stack is a normal stack of 256 values.
2023-03-26 13:17:21 <nature      > here is the base words: https://git.blazebone.com/pochi/tree/pochi.c#n143
2023-03-26 13:44:46 <MrMobius    > what does this run on?
2023-03-26 13:45:10 <thrig       > blood, sweat, and tears, same as any other software
2023-03-26 13:48:47 <Zarutian_iPad> tears of engineers, sweat of overworked logisticans, and the blood of the cheap labour assembling it
2023-03-26 13:55:41 <nature      > x86
2023-03-26 13:55:57 <nature      > (and some blood and tears too)
2023-03-26 13:56:30 <nature      > I develop on OpenBSD, should compile on Linux too
2023-03-26 13:56:34 <KipIngram   > Hi nature!
2023-03-26 13:56:35 <Zarutian_iPad> so, out dated legacy ISA then
2023-03-26 13:56:43 <KipIngram   > Happy weekend.
2023-03-26 13:56:44 <nature      > and I plan to port it to wasm as well
2023-03-26 13:57:02 <nature      > 'sup KipIngram!
2023-03-26 13:57:02 <Zarutian_iPad> RISC is where it is at!
2023-03-26 13:57:11 <Zarutian_iPad> RISC-V that is
2023-03-26 13:58:31 <nature      > It's writtend in C, if there is a C compiler and SDL2 on RISC-V it should be relatively easy to have it running there, I have around 5-8K loc with the full graphical environment.
2023-03-26 13:59:28 <nature      > But yeah, basically just some feedback/guidance on the choice of base set of words would be much appreciated as I still consider myself quite the forth neophyte
2023-03-26 14:03:50 <Zarutian_iPad> depends, lot of these seemto be to speed up things, no?
2023-03-26 14:07:36 <Zarutian_iPad> the primitives I have got in my FCPU-16 thing are: NOP UM+ AND XOR 1LBR 1+ @ ! DUP DROP SWAP SKZ >R R> EXT/EFI EXIT
2023-03-26 14:08:26 <Zarutian_iPad> but then again FCPU-16 is a conventional dual stack machine based with I/O in memory address space
2023-03-26 14:09:12 <Zarutian_iPad> bbl
2023-03-26 14:12:11 <nature      > yeah, I initially started with with 5 bits opcodes (the 32 from the F18a) but I thought that things like "/" and "swap" could quickly become quite slow and wouldn't really take advantage of the fact that it's a vm and not running on an actual f18.  So I bumped the opcodes to 6 bits and found myself with lots of room to add opcodes that would be quite convenient/faster. But I am really not sure if I should still keep my opcodes count low
2023-03-26 14:12:11 <nature      > (for simplicity) or just go ahead and make full use of 64 opcodes as well.
2023-03-26 14:12:45 <nature      > wow that's low, and what is skz?
2023-03-26 14:13:20 <nature      > Also any opinions on using a and b registers for the @ and ! words?
2023-03-26 14:14:10 <thrig       > sounds like it needs some A/B testing
2023-03-26 14:23:12 <Zarutian_iPad> 
ACTION is back

2023-03-26 14:23:35 <Zarutian_iPad> skz is SKip if Zero
2023-03-26 14:24:03 <Zarutian_iPad> skips the next instruction cell
2023-03-26 14:46:09 <Zarutian_iPad> 
ACTION sees https://pm.stackexchange.com/questions/34768/why-are-developers-expected-to-estimate-tasks-at-all and notes that these kind of “estimates” requests are really promise extraction

2023-03-26 14:47:08 <Zarutian_iPad> promise extraction is kind of toxic behaviour that should never be allowed to go on
2023-03-26 14:47:52 <thrig       > didn't the Romans take hostages for when the celts revolted?
2023-03-26 14:56:24 <MrMobius    > nature, this type of consideration is actually a bit complicated. trying to make the opcodes as small as possible might not pay off at all and saving a handful of bytes doesn't do much on x86
2023-03-26 14:57:06 <MrMobius    > also, if you are developing the forth to develop another program, just add whatever words you need as you need them
2023-03-26 14:57:42 <MrMobius    > you will probably lose a lot of performance but don't worry too much about that. it probably doesn't matter for most things
2023-03-26 15:01:35 <thrig       > whipitup Y is very different from long term support of X
2023-03-26 15:31:25 <nature      > Got it, thanks MrMobius
2023-03-26 18:55:01 <KipIngram   > Zarutian_iPad: I mostly agree.  In some cases a lead engineer with good experience can make a good guess at the duration of a project, and that can help business peeps make rough plans.  But it really shouldn't ever be looked upon as a guarantee or a requirement.  It's impossible to know for sure how work will go until you do it.
2023-03-26 18:55:23 <thrig       > oh boy crunch time
2023-03-26 18:55:35 <KipIngram   > If management typically treats such things as requirements, then the dev team will learn to sandbag.
2023-03-26 18:55:45 <KipIngram   > So the bosses don't get the best prediction.
2023-03-26 18:56:15 <KipIngram   > And the bosses learn to counter that by demanding "less than the estimate."
2023-03-26 18:56:20 <thrig       > or you end up with what some have called "softare death marches"
2023-03-26 18:56:39 <KipIngram   > What should be a cooperative relationship becomes an adversarial game.
2023-03-26 18:58:03 <KipIngram   > It really does seem incredibly common for executives to operate as though they can simply will something into existence.
2023-03-26 19:02:00 <thrig       > stacked rankings probably didn't help with the adversarial thing
2023-03-26 20:33:29 <KipIngram   > Well, I think there's a lot wrong with all those "big company process" type things.
2023-03-26 20:33:51 <KipIngram   > My most satisfying days career wise were in a company with barely over a hundred employees.
2023-03-26 20:34:45 <KipIngram   > Felt like I really made a difference there.  Not that someone else couldn't have done the job, but at that time it was me they looked to to do it, and I took that really seriously.
2023-03-26 20:35:26 <KipIngram   > Sometimes I think I'm probably lucky I'm still married - I overworked quite a lot those years.  And my wife was quite young at the time, and hadn't had her own "high pressure work" experience yet.
2023-03-26 20:36:06 <KipIngram   > Then when she did later, it was a lot easier for me to be understanding about it, since I recognized exactly what she was going through.
2023-03-26 20:37:18 <KipIngram   > Sometimes I miss it, but then I think, "Do I REALLY want to do all that again?"