Why are the HP Prime and HP-35s so unpopular?

[Peet]

Both have at least as many or even more pluses than disadvantages compared to the popular 50G and 42S calcutators.
The current HP-17bII+ also seems to be a pretty good calcutator in the sum of its properties, but it hardly gets any positive mention.

From my point of view, the lack of external storage is the biggest weakness of the 35s but this is exactly what it shares with the 42s.

The prime, on the other hand, even seems to be superior to the 50g in every aspect.

What makes the 42s/50g so good and the 35s/Prime so bad?

Are these calcutators really so bad that it is worth dreaming about the past instead of wishing for improvements?

[Maximilian Hohmann]

Hello!

This is mainly a matter of personal taste I think rather than objective criteria.

(01-22-2021 01:24 PM)Peet Wrote:  The prime, on the other hand, even seems to be superior to the 50g in every aspect.

I second that. The only thing I don't like about the Prime is the color scheme of the keyboard. For the sake of a complete collection I bought an HP-50g some years ago. Put batteries inside, played with it for half an afternoon, removed the batteries and never touched it again. This calculator has exactly zero appeal to me and not only becauso of RPL...

(01-22-2021 01:24 PM)Peet Wrote:  What makes the 42s/50g so good and the 35s/Prime so bad?

I never disliked the 35s at all. It came on the market when writing large programs for programmable calculators already was a thing of the distant past (for professional use at least), so the lack of mass storage or the "dreadful bugs" do not really affect the casual user. Apart from it's battery drain (I have three or four or fife (don't really know) and all of them drain a set of batteries within two or three months even if they are never turned on) there is nothing really wrong with it in my view, at least not at the price it was sold for.
On the other hand I don't really like the 42s (as I never liked the 41C either). Accessing functions through "XEQ Alpha <some three letter function code that I don't really want to remember>" is not how I like to use a calculator, nor are the menus.

But again, it is purely a matter of what one is used to and what one's personal preferences are.

Regards
Max

[Dave Britten]

The Prime is a vast departure from old designs, and lacks a lot of the customization features that we're accustomed to, particularly custom soft-key menus from the 48. Also its RPN mode is fairly crappy. I'm kind of surprised nobody has written a Nut emulator/15C simulator for it.

The 35S has more bugs than a roach motel. It's so close to being great, and the hardware design is lovely, but the software is such a mess.

[Howard Owen]

(01-22-2021 01:24 PM)Peet Wrote:  Are these calcutators really so bad that it is worth dreaming about the past instead of wishing for improvements?

Speaking for myself, I'm enamored by the old machines for more than their utility as calculators. It's not just the build quality, or snappy keyboard "feel." I learned to program on an HP41C. I love bringing that old technology into the 21st century. For me it's nostalgia, and a sense of fun in playing with the old machines.

This is probably not the right forum if you are looking for 35S or Prime enthusiasts. Have you tried the Prime forum? You might get different answers to your question there.

[ttw]

It's a matter of preference. In my case, the Prime had two big problems. First, it doesn't do RPL or RPN or the like; if I want to write out formulas, I'll just use Silverfrost Fortran which is much faster and easier to use. Second, when I tried to update the Prime using the USB, it shorted the motherboard on a desktop; thus I found it cheaper not to use the Prime.

It was neither user-friendly (with me as the user), nor hardware friendly to being updated.

[Eddie W. Shore]

The HP Prime is (roughly?) based off the HP 38G/39G calculators, which are not quite as popular as the 28/48/49/50 series. As for the RPN, I never use the Prime's RPN mode, it's serviceable as a basic RPN mode, like the HP 21 or HP 45, except with an infinite stack.

The CAS is a separate mode from numeric mode, and any integration between two is at best rough.

As for me, the Prime is one of the calculators I use most often and it is one my favorites.

There are two criticisms of the HP 35S:
* Lack of Polar/Rectangular conversions in its function set
* Many bugs that were not addressed

Giving the 35S the ability to transfer programs to offline storage via USB would have helped.

[dxaren]

35s:
The lack of direct P<->R conversion!!!!
The angle symbol that looks like the figure 8.
The display presentation of complex and imaginary values.
The strange behaviour of checksum for program.
Battery life.

/ in daily use: 15LC, 41CV and 35S
/ seldom use: 19BII, 29C, DM16,

[Eddie W. Shore]

(01-22-2021 02:26 PM)dxaren Wrote:  35s:
The lack of direct P<->R conversion!!!!
The angle symbol that looks like the figure 8.
The display presentation of complex and imaginary values.
The strange behaviour of checksum for program.
Battery life.

/ in daily use: 15LC, 41CV and 35S
/ seldom use: 19BII, 29C, DM16,

Thanks for reminding me: the lack of Real and Imaginary part functions for complex numbers.

[Marco Polo]

(01-22-2021 01:24 PM)Peet Wrote:  Both have at least as many or even more pluses than disadvantages compared to the popular 50G and 42S calcutators.
The current HP-17bII+ also seems to be a pretty good calcutator in the sum of its properties, but it hardly gets any positive mention.

From my point of view, the lack of external storage is the biggest weakness of the 35s but this is exactly what it shares with the 42s.

The prime, on the other hand, even seems to be superior to the 50g in every aspect.

What makes the 42s/50g so good and the 35s/Prime so bad?

Are these calcutators really so bad that it is worth dreaming about the past instead of wishing for improvements?

I think is more a matter of personal tastes, as others stated, but also of the typical use that a user makes of his calculators.
In my case, the Prime is a step backwards compared to the 50g as i like RPL and my use involves heavy use of units of measure in calculation and solver and the user of user keyboard and custom menus.
So i am stuck on 48/50 calculators and for me the 50g is better than the Prime.

For others, the Prime is better than 50g and i am sure they are as right as i am.
:-)

[Peet]

(01-22-2021 01:41 PM)Maximilian Hohmann Wrote:  This is mainly a matter of personal taste I think rather than objective criteria.

(01-22-2021 02:58 PM)Marco Polo Wrote:  I think is more a matter of personal tastes, ...

If that were the case, then everything would be OK, but mostly I have a different impression. I have read negative comments on the Prime or 35s in many posts (often without any explanation), but I've never noticed that with the 50g or 42s. On the other hand, many hymns to the 50g an 42s, not so much to the 35s.

Even in this thread only disadvantages to the Prime or 35s were named, none to the 50g/42s. It almost seems as if the 35s and Prime are not recommended and the 42s and 50g are nearly perfekt and error-free.

For my personal needs the 35s and 42s are equal as daily used calculators and even 50g and Prime would be similar replacements for my 48g(+).

[Massimo Gnerucci]

(01-22-2021 02:21 PM)Eddie W. Shore Wrote:  As for the RPN, I never use the Prime's RPN mode, it's serviceable as a basic RPN mode, like the HP 21 or HP 45, except with an infinite stack.

Really? I was under the impression that it was a fixed 128 levels stack.

[Dave Britten]

(01-22-2021 02:21 PM)Eddie W. Shore Wrote:  The CAS is a separate mode from numeric mode, and any integration between two is at best rough.

Oh yeah, this always bugged me. The TI-92/89, Nspire, Casio Classpad, HP 28/48/49/50, etc. have no such artificial division between numeric and CAS operation. The fact that the Prime separates these modes and even gives them some differing syntax and variable name conventions is baffling.

[Sylvain Cote]

I see these types of HP calculators:

• RPN handheld calculators
  → began as a electronic slide rules replacement and then evolved from that
  → integrated system
  → closed platform
  → no I/O (exception: HP-97S)
  → programmable: yes on some models, keystrokes programming
  → customer type (in order): engineer, financial, math & student
  
• RPN handheld computers (41)
  → integrated system
  → open-ended platform
  → full I/O (HP-IL, peripherals & modules)
  → programmable: yes, advanced keystrokes programming
  → customer type (in order): engineer, financial, math & student
  
• BASIC handheld computers (71/75)
  → integrated system
  → open-ended platform
  → full I/O (HP-IL, card reader & modules)
  → programmable: yes, higher level languages (BASIC & FORTH)
  → customer type (in order): engineer & math
  
• RPN over System RPL calculators
  → integrated system
  → closed platform
  → no I/O (exception: infra-red output on some models)
  → programmable: yes, solver, keystrokes programming
  → customer type (in order): student, math, engineer & financial
  
• AOS over System RPL calculators
  → alternative to RPN over System RPL calculators
  → integrated system
  → closed platform
  → no I/O (exception: infra-red output on some models)
  → programmable: yes, solver, keystrokes programming?
  → customer type (in order): student, math, financial & engineer
  
• RPL calculators
  → bring symbolic math to the calculator world
  → integrated system
  → open-ended platform
  → limited I/O (serial)
  → programmable: yes, higher level language
  → customer type (in order): student, math & engineer
  
• PPL calculators
  → applets based system
  → limited or no integration between apps
  → open-ended platform
  → programmable: yes, higher level language
  → customer type: student
  
• Rebranded AOS/RPN calculators
  → no comments
  
  

So, peoples needs and preferences will favor one type over the other.

In my case, when I ...

• need I/O → I use: 41, 71 & 75
• do embedded programming → I use: 16C, 41+module
• do scientific calculations → I use: 41, 32S, 25C, etc
• need symbolic math → I use: 48GX & 50G
  

I prefer integrated system (with full I/O when possible) over app based ones.

Sylvain

[Marco Polo]

(01-22-2021 03:28 PM)Peet Wrote:  If that were the case, then everything would be OK, but mostly I have a different impression. I have read negative comments on the Prime or 35s in many posts (often without any explanation), but I've never noticed that with the 50g or 42s. On the other hand, many hymns to the 50g an 42s, not so much to the 35s.

Even in this thread only disadvantages to the Prime or 35s were named, none to the 50g/42s. It almost seems as if the 35s and Prime are not recommended and the 42s and 50g are nearly perfekt and error-free.

For my personal needs the 35s and 42s are equal as daily used calculators and even 50g and Prime would be similar replacements for my 48g(+).

You should consider that the forum is mainly populated by long term HP calcs users and (this is true especially for 50g), the 42s and 50g are the top edge of the development of RPN and RPL calculators.
They both have bugs, but are very well known and in imho don't impact the everyday usability.
I don't know well the 42s, but i can say that the 50g has things i don't like if compared to the 48gx: i prefer the 48 screen header, the 48 has better layout, the 7-level stack screen should be optionally rearranged as 5-level with large font, the 50g menu are somewhat confused.

The 35s, mainly due to its external design, imho generates high expectations but unfortunately bugs and some strange "flaws" (the display, for example) did not satisfy the users. This is a real pity as the keyboard is pretty decent and the body is robust. A better firmware would have helped. I own one since 2008, but never fell in love with it. Maybe i cannot get used to RPN vs RPL stack paradigma (stack lift....).

I also tried the Prime as emulator, but gave up after some days: its design, with different enviroment for each task, the difficulty of passing variables to other environments, the lack of unit of measures in 48-like way are keeping me away.
In my opinion the 50g is the last HP calc aimed to engineers AND matematicians AND partially students, while the Prime dropped the engineers.

[aurelio]

I like particularly Sylvain's and Marco Polo answers, for each job its dedicated tool if available.
I like as Marco the 48's screen and agree when he writes "the 48 screen header, the 48 has better layout, the 7-level stack screen should be optionally rearranged as 5-level with large font, the 50g menu are somewhat confused."
I have no more the prime, i found its keyboard really difficult to read ...
I never had a 35s, it's a nice calculator hestetically but never I bought it , nor used, so no personal experience, but I read quite about its various bugs.
I don't think it's a matter for saying a calculatore is bad or good, it's a matter of taste and usage

[Valentin Albillo]

Peet Wrote:Why are the HP Prime and HP-35s so unpopular?

Many replies here have stated more or less that "it's a matter of taste and usage".

True. But there's another factor, which for me is the essential one: it's a matter of trust.

I can trust all the classic models, from the HP-21 to the HP-42S and of course the HP-71B. They do have their share of bugs but they're well known and well documented, none of them are catastrophic and most are very rarely encountered, if at all. In practice this means that you can use them confidently, without the machines resetting unexpectedly once and again for no known reason, and with the peace of mind that comes with knowing you won't suddenly lose your programs to a nasty bug and even better, that barring physical *hardware* malfunctions your programs will continue to run over time as they always did, producing the exact same results in a deterministic way. In short: you can trust the machine.

That desirable state of affairs doesn't apply to either the Prime or the HP-35S. The latter not only has a very ill-devised and incomplete intruction set, with abysmal support for complex numbers and vectors, very basic and essential functionality sorely missing, and incredible inefficiencies everywhere (such as numbers in program lines being treated as if they were equations (!!), which slows down execution by 10x or more if used within a loop), but additionally has lots of bugs, some of which do cause the loss of all your painfully entered programs which thus must be re-entered and re-checked anew, and others do *change* some code in your programs causing them to perform incorrectly, which you might notice or (much worse) not. You can't trust the results and after being bitten several times by this you get fed up and simply put out the beastie in a drawer for good.

As for the former, the Prime, same tune. It has zillions of bugs already reported and in all probability there are zillions of bugs yet to discover, so again the question arises: Can you trust the results you get ? Can you trust that your programs will run the same and produce the same results after the next firmware update ?.

My answer to these questions is "No and No". I would never trust for serious work the results returned by the Prime without redoing them in a wholly different machine and checking if they do match. And if I have to do that to get peace of mind I might as well use a trusted machine from the get go and give the Prime a miss.

Worse still, after fighting the known bugs and the picky syntax of the miriad functionalities and their many caveats and inconsistencies, you may have your long, complicated technical Prime program finally running fine and thoroughly checked, then sooner or later an update internally changes something or reintroduces/creates bugs while attempting to correct other bugs, and without warning your program now either doesn't compile/run or much worse, produces incorrect results which might easily go unnoticed and may take you lots of time to find and debug only to discover that it wasn't your fault after all but this or that update messing things up.

The bottom line: you can't trust the results for serious work.

And as I said, it's a matter of trust.

V.

[toml_12953]

(01-22-2021 04:05 PM)Dave Britten Wrote:  

(01-22-2021 02:21 PM)Eddie W. Shore Wrote:  The CAS is a separate mode from numeric mode, and any integration between two is at best rough.

Oh yeah, this always bugged me. The TI-92/89, Nspire, Casio Classpad, HP 28/48/49/50, etc. have no such artificial division between numeric and CAS operation. The fact that the Prime separates these modes and even gives them some differing syntax and variable name conventions is baffling.

Which is why I never use CAS mode. Even so, I get messed up sometimes with inadvertent interaction between the two.

[Liam Hays]

I learned to use RPN and about HP calculators in general from an HP-32S, a wonderful little machine. Compared to that, the 35s is vastly more powerful in terms of capability---complex support beyond just arithmetic, vectors, equation and system solving in a natural way, a clear and readable two-line screen, a significantly more advanced version of the 32S' programming language, and enough memory to put all that to good use.

And therein lies the problem. The 35s is so darn close to being a nearly perfect machine. In my opinion, if HP had done these three things, bugs like the checksum bug, cosine bug, and self-test problems would be nothing more than an inconvenience:

1. Fixed the keyboard. When I pulled by 35s out of its packaging, the R/S, GTO, and XEQ keys already looked like they had been used for years, and just days later, the right-shift key decided it needed extra force to register. While the actual keypress feel of the keyboard is fine, to me, fixing the actual key sensors should have been number one on HP's list.
2. Made complex support more than half-baked. Having a dedicated i key is fine, but if I can't take a square root of a negative number and get a complex answer out, I would rather have STO and RCL as dedicated keys instead. (On that topic, using the final layout, they could have at least made the MODE key shifted, and moved STO to its own key).
3. Finally, given the poor thing a processor greater than the 6502 or whatever is in there. A few months ago, I tested the integration speed between the 35s and the 32S, using the example from the 32S, and the 32S finished in three times the speed of the 35s. The kind of math operations the 35s supports demands speed, which is readily available today and probably was in 2007.
   

On the topic of speed, this page ranks the performance of the 35s at 9.5, only about three times greater than the HP-41, released 28 years before (though comparing the 35s to the 41C is a completely different ballgame). All in all, I think that the HP-35s is a good calculator, but the problems it suffers from are severely debilitating. This is really a shame, since it's probably the last standard scientific calculator we'll ever get from HP.

Anyway, I'll be quiet now. PS: this is my first post here!

[Valentin Albillo]

(01-23-2021 12:03 AM)Liam Hays Wrote:  Anyway, I'll be quiet now. PS: this is my first post here!

Welcome to the forum. Lots of interesting things here and a few dullards as well.

Would you care to post the exact integral you used to compare integration speeds ?

V.

[trojdor]

(01-23-2021 12:03 AM)Liam Hays Wrote:  A few months ago, I tested the integration speed between the 35s and the 32S, using the example from the 32S, and the 32S finished in three times the speed of the 35s.

Although I don't have access to the original 32s, I recently compared the real world integration of the cumulative normal distribution function (CDF) between a 32Sii and the 35s, and saw a much smaller difference....although the 32sii WAS slightly faster.

Both functions were entered as equations, then integrated.
The 32sii finished in about 7.25 seconds, and the 35s finished in about 10.25 seconds....about 40% longer time.

Admittedly, it would indeed be nicer to have a faster processor in the 35s.