Collection of wishes for future SwissMicros products - discussion following Bob Prosperi's HHC 2022 talk

General discussion about calculators, SwissMicros or otherwise
ben.titmus
Posts: 49
Joined: Wed Feb 24, 2021 7:50 pm
Location: Cambridge, UK

Re: Collection of wishes for future SwissMicros products - discussion following Bob Prosperi's HHC 2022 talk

Post by ben.titmus »

Walter wrote:
Mon Oct 10, 2022 6:39 pm
... and its logical but weird way of addressing registers - postfix for postfix' sake
This certainly makes everything very consistent, but it does take some getting used to. It's much like programming in Forth. Forth is an amazingly simple language and can be implemented in very little code. For example in RPN we have to cope with the buffer on RCL which I'm sure is more complicated to implement than if we just took the argument off the stack. With more complicated programs, if you have a reusable 'function' that needs to take as an argument which register to load from, you end up needing to use indirect registers. The WP43 makes this much easier by allowing any register to be indirect, but it's still a bit of extra work. So I think for reusable components RPL manages much better. Not so much for short simple programs where it ends up being very verbose.
Walter wrote:
Mon Oct 10, 2022 6:39 pm
I can grab any RPN calc and start working immediately while I continuously make errors on RPL calcs. Although RPL looks like being (or even is) more coherent and powerful, I think RPN is suited better for quick and dirty solving real-world problems. YMMV
I'd agree that RPN is better suited for quick and dirty problem solving. But I think RPL programs are easier to reason about and write than long RPN keystroke programs.
packrat
Posts: 14
Joined: Thu Jul 21, 2022 4:12 am

Re: Collection of wishes for future SwissMicros products - discussion following Bob Prosperi's HHC 2022 talk

Post by packrat »

As someone who wrote a bunch of smaller programs mostly on an HP32sII and then later wrote a bunch of stuff on an hp48GX, I think that RPN keystroke programming is great for automation of a thing that starts out as a process you might do yourself. They can absolutely get complex and do interesting things to values on a stack.

However, if you're going to be writing little fragments of code that become a library of tools you can more easily construct large things conceptually from, I found RPL on a 48 was just considerably easier to use in that fashion. In at least one engineering exam that required a blanker calculator, it was efficient to simply recreate the useful tools in the exam on a 48.

This may even be an engineering vs computer science perspective difference.

B>
packrat
Posts: 14
Joined: Thu Jul 21, 2022 4:12 am

Re: Collection of wishes for future SwissMicros products - discussion following Bob Prosperi's HHC 2022 talk

Post by packrat »

Oh, and FWIW, the ability to see larger stretches of a program on a DM42 screen in the editor I think exhances RPN keystroke programming significantly and the larger memory/functiony definitions close that gap.

I find it hard to express the degree to which the larger window of keystroke/RPN programming available on a DM42/DM41X makes a difference to my relationship with those programs.
ben.titmus
Posts: 49
Joined: Wed Feb 24, 2021 7:50 pm
Location: Cambridge, UK

Re: Collection of wishes for future SwissMicros products - discussion following Bob Prosperi's HHC 2022 talk

Post by ben.titmus »

packrat wrote:
Tue Oct 11, 2022 12:20 pm
... I think that RPN keystroke programming is great for automation of a thing that starts out as a process you might do yourself.
...
However, if you're going to be writing little fragments of code that become a library of tools you can more easily construct large things conceptually from, I found RPL on a 48 was just considerably easier to use in that fashion.
This is a very good way of putting it, and expresses what I was wanting to say.
packrat wrote:
Tue Oct 11, 2022 12:31 pm
... the ability to see larger stretches of a program on a DM42 screen in the editor I think exhances RPN keystroke programming significantly and the larger memory/functiony definitions close that gap.
I think the UI additions to the DM42 and even more so the WP43 really make the difference between RPN and RPL very clear. There are some differences to how the stack functions, and larger differences in the programming paradigm. But they are both very easy to program and use now.

My ideal calculators are the WP43 and a newRPL calculator. We're not that far away from the WP43 physical calculator. The newRPL calculator will be more challenging. I currently run it on my HP-50G; I have a G2 Prime (so it won't run on that). Really newRPL is lacking some features (although most of the programming functionality is implemented and works really well) but most of all it lacks a hardware platform. The HP models it works on are not being produced anymore. So my ideal next SM calculator would be a newRPL platform. But that would be a lot of work!
redglyph
Posts: 177
Joined: Sat Dec 22, 2018 11:45 am

Re: Collection of wishes for future SwissMicros products - discussion following Bob Prosperi's HHC 2022 talk

Post by redglyph »

ben.titmus wrote:
Tue Oct 11, 2022 1:45 pm
I think the UI additions to the DM42 and even more so the WP43 really make the difference between RPN and RPL very clear. There are some differences to how the stack functions, and larger differences in the programming paradigm. But they are both very easy to program and use now.

My ideal calculators are the WP43 and a newRPL calculator. We're not that far away from the WP43 physical calculator. The newRPL calculator will be more challenging. I currently run it on my HP-50G; I have a G2 Prime (so it won't run on that). Really newRPL is lacking some features (although most of the programming functionality is implemented and works really well) but most of all it lacks a hardware platform. The HP models it works on are not being produced anymore. So my ideal next SM calculator would be a newRPL platform. But that would be a lot of work!
RPN is funny but not really intuitive, even in its last iteration, because you have to know when the stack is lifted and when stack lifting is disabled. RPL is much simpler and the programs are easier to understand, so I would easily sacrifice the rare benefit of RPN for the sake of clarity (just my opinion and experience using it).

I think newRPL is still in beta, so that would be a problem. I haven't tried it, only watched a short presentation about it.
User avatar
Walter
Posts: 3070
Joined: Tue May 02, 2017 11:13 am
Location: On a mission close to DRS, Germany

Re: Collection of wishes for future SwissMicros products - discussion following Bob Prosperi's HHC 2022 talk

Post by Walter »

redglyph wrote:
Tue Oct 11, 2022 6:58 pm
RPN is funny but not really intuitive, even in its last iteration, because you have to know when the stack is lifted and when stack lifting is disabled. RPL is much simpler and the programs are easier to understand, so I would easily sacrifice the rare benefit of RPN for the sake of clarity (just my opinion and experience using it).
Seasoned programmers know ASL by heart. See above once again.
WP43 SN00000, 34S, and 31S for obvious reasons; HP-35, 45, ..., 35S, 15CE, DM16L S/N# 00093, DM42β SN:00041
Linus_Sch
Posts: 60
Joined: Sun May 09, 2021 9:33 am

Re: Collection of wishes for future SwissMicros products - discussion following Bob Prosperi's HHC 2022 talk

Post by Linus_Sch »

ben.titmus wrote:
Mon Oct 10, 2022 12:55 pm
Personally I'd really like a SwissMicros RPL platform. However, this would be a significant difficulty since it would need to be much more powerful and the case would need to completely change (as well as needing better batteries). Then there's the question of the display - should it be colour (I would rule out touchscreen which I found a pointless 'feature' on the Prime)?
The hardware of the DMCP, when powered by USB, is actually more powerful than that of the 50g. On battery the speed is about a third, which IMO would not be a dealbreaker for newRPL, but I have not tested newRPL extensively yet. If the tidbit from Bob's presentation that the new version is as fast on battery as the old one on USB power then we're in great shape on processing power.

However, it might need more RAM, if we want newRPL running as an application within the DMCP OS. The SoC does have provisions for external extensions of memory.

I would go for pretty much the same thing as the DM32 hardware. Two extra rows of keys (one extra row would be same as 48, but that is two keys fewer than the 50g, and I like keys), extended RAM. Maybe a somewhat higher power battery solution, enabling a bit of extra speed, but please stay with long life coin cells. Ideally I'd like some solution for interfacing with other stuff as well, but that might up the battery requirements too much.
ben.titmus wrote:
Tue Oct 11, 2022 1:45 pm
My ideal calculators are the WP43 and a newRPL calculator. We're not that far away from the WP43 physical calculator. The newRPL calculator will be more challenging. I currently run it on my HP-50G; I have a G2 Prime (so it won't run on that). Really newRPL is lacking some features (although most of the programming functionality is implemented and works really well) but most of all it lacks a hardware platform. The HP models it works on are not being produced anymore. So my ideal next SM calculator would be a newRPL platform. But that would be a lot of work!
Above I make the argument that SM hardware for newRPL need not be a lot of work. Indeed, while quite compromised in terms of key layout, newRPL could run quite well on current SM hardware. However: it would be a lot of work.

Software work. And decisions, the first one being whether to make a version of newRPL that strips out all the OS stuff and runs on the DMCP OS instead (which might be a whole lot of work or might not, depends on the software architecture of newRPL, and might need more RAM as noted above but might also slim down enough in the process to not really need it) - or to port newRPL including its OS functionality to the SM hardware (which might also be a whole lot of work, depending more on how large the hardware differences are).

Finally, newRPL needs a program editor and an equation editor and solvers and libraries and CAS. I'm not sure how far along that journey is as of now. Does the original HP equation library and periodic table run on newRPL, is it that compatible? I do think progress would be faster if the hardware was made and launched, "for developers only" at first, maybe don't show it on the main site and just link to the product page from this here forum or something.

Then there is the final wrench in this: my optimal RPL calculator may not be newRPL. It may be a hardware platform with an emulator that can run the available HP ROMs. I'm not sure, it depends on how complete and how compatible newRPL can be.
Linus_Sch
Posts: 60
Joined: Sun May 09, 2021 9:33 am

Re: Collection of wishes for future SwissMicros products - discussion following Bob Prosperi's HHC 2022 talk

Post by Linus_Sch »

Walter wrote:
Tue Oct 11, 2022 7:55 pm
redglyph wrote:
Tue Oct 11, 2022 6:58 pm
RPN is funny but not really intuitive, even in its last iteration, because you have to know when the stack is lifted and when stack lifting is disabled. RPL is much simpler and the programs are easier to understand, so I would easily sacrifice the rare benefit of RPN for the sake of clarity (just my opinion and experience using it).
Seasoned programmers know ASL by heart. See above once again.
Sure, but this line of discussion did start with the learning curve. And the fact that quick-to-grasp, from a starting point of no previous experience with either RPN or RPL, can help attract more new customers.

I find the following to be true with curious young people in any field, about anything: if they can understand and start to utilize the power within a few minutes of getting their hands on it, they tend to want it! If half an hour is needed the level of curiousity needed for someone to give it a fair attempt is much higher.
redglyph
Posts: 177
Joined: Sat Dec 22, 2018 11:45 am

Re: Collection of wishes for future SwissMicros products - discussion following Bob Prosperi's HHC 2022 talk

Post by redglyph »

Walter wrote:
Tue Oct 11, 2022 7:55 pm
redglyph wrote:
Tue Oct 11, 2022 6:58 pm
RPN is funny but not really intuitive, even in its last iteration, because you have to know when the stack is lifted and when stack lifting is disabled. RPL is much simpler and the programs are easier to understand, so I would easily sacrifice the rare benefit of RPN for the sake of clarity (just my opinion and experience using it).
Seasoned programmers know ASL by heart. See above once again.
Exactly, seasoned macro programmers, that's my point. I see no contradiction there.
ben.titmus
Posts: 49
Joined: Wed Feb 24, 2021 7:50 pm
Location: Cambridge, UK

Re: Collection of wishes for future SwissMicros products - discussion following Bob Prosperi's HHC 2022 talk

Post by ben.titmus »

Linus_Sch wrote:
Tue Oct 11, 2022 8:56 pm
The hardware of the DMCP, when powered by USB, is actually more powerful than that of the 50g. On battery the speed is about a third, which IMO would not be a dealbreaker for newRPL, but I have not tested newRPL extensively yet... However, it might need more RAM, if we want newRPL running as an application within the DMCP OS. The SoC does have provisions for external extensions of memory.
I can't agree with this assessment. Even if the clockspeed were to be sufficient to be the same as the 50G that would mean that we're as good as a really old calculator. The DM42 is significantly faster than the HP-42S and we'd expect a newer RPL calculator to be in the same league as the Prime, not the 50G. Whilst the SoC allows external memory, that memory is likely to be slow in comparison (I haven't checked the specs though). In any case, if an extra row of keys is needed then you need a different PCB, and given all the associated costs of creating a new PCB design, you might as well change the SoC at the same time.
Linus_Sch wrote:
Tue Oct 11, 2022 8:56 pm
I would go for pretty much the same thing as the DM32 hardware. Two extra rows of keys (one extra row would be same as 48, but that is two keys fewer than the 50g, and I like keys), extended RAM. Maybe a somewhat higher power battery solution, enabling a bit of extra speed, but please stay with long life coin cells. Ideally I'd like some solution for interfacing with other stuff as well, but that might up the battery requirements too much.
As I say, I think you'd need a better SoC. If you add extra batteries then you have two choices: move to AAA or add extra coin cells. At that size I'd prefer AAA, which is the same decision HP made (for good reason). (I'm not really sure why anyone would prefer coin cells except because it keeps the size down, in which case surely the size is the requirement rather than the batteries?) Nowadays the only interface that makes sense is USB-C which the DM32 will have. Maybe IR if you have an old printer.
Linus_Sch wrote:
Tue Oct 11, 2022 8:56 pm
... it would be a lot of ... software work. And decisions, the first one being whether to make a version of newRPL that strips out all the OS stuff and runs on the DMCP OS instead.
I've only briefly looked at the newRPL source a while ago, but I think it's fairly OS/platform agnostic, in which case it would be easy to port to DMCP.
Linus_Sch wrote:
Tue Oct 11, 2022 8:56 pm
Finally, newRPL needs a program editor and an equation editor and solvers and libraries and CAS. I'm not sure how far along that journey is as of now. Does the original HP equation library and periodic table run on newRPL, is it that compatible? I do think progress would be faster if the hardware was made and launched, "for developers only" at first, maybe don't show it on the main site and just link to the product page from this here forum or something.
This is a chicken and egg situation. But I think it would be unlikely that SM would take such a risk with the software in it's current state. There hasn't been any development on newRPL since the beginning of this year. However, I think most further development would be in RPL so anyone who knows RPL well enough could join in. If I remember correctly I don't think they are expecting to implement CAS.
Linus_Sch wrote:
Tue Oct 11, 2022 8:56 pm
Then there is the final wrench in this: my optimal RPL calculator may not be newRPL. It may be a hardware platform with an emulator that can run the available HP ROMs. I'm not sure, it depends on how complete and how compatible newRPL can be.
The advantage of newRPL is that it's open source so it can continue to be developed. I'm not particularly interested in getting a SM 50G, because I have two 50Gs already (although both running newRPL).
Post Reply