About the 39GS

[toml_12953]

(02-18-2018 12:42 PM)Gerald H Wrote:  

(02-18-2018 12:32 PM)toml_12953 Wrote:  What makes it so horrible? Its programming language is the same as the Prime which I find the best language of any calculator I've ever used (I've used most programmables from the HP 9100A and later)

Exactly! & that reveals the underlying grotesqueness of the Prime.

What would you use? That horrible write-only language where everything is in
« » ? RPL is good for one-off calculations and I use it for those. I can see the attraction of building functions like layers of an onion. For permanent programs, though, give me a self-documenting language where I can go back and immediately see how I did something in a program written months or even years ago. People may say that RPL is closer to the machine and therefore more efficient. I say that nowadays, the hardware is advanced enough to make that a moot point. My time is most important. I want the machine to do the busy work of parsing, hierarchy, etc..

[Gerald H]

I don't know, toml_12953, how to answer you.

Perhaps an example:

If you look at this thread

http://www.hpmuseum.org/forum/thread-10026.html

you will see the development of a programme that I enjoyed working on, in the hope of producing a short & fast programme in an environment with, by today's standards, limited resources (a 50g).

The final version is able to find a partition of a 44,400 digit number in c 21 sec, which I consider reasonably fast.

True, more modern hardware will deal with the problem much more efficiently.

So perhaps you, or some interested Prime adept, could write a Prime programme to partition a 44,444 digit integer in palindromic parts more efficiently?

[toml_12953]

(02-18-2018 07:23 PM)Gerald H Wrote:  I don't know, toml_12953, how to answer you.

Perhaps an example:

If you look at this thread

http://www.hpmuseum.org/forum/thread-10026.html

you will see the development of a programme that I enjoyed working on, in the hope of producing a short & fast programme in an environment with, by today's standards, limited resources (a 50g).

The final version is able to find a partition of a 44,400 digit number in c 21 sec, which I consider reasonably fast.

True, more modern hardware will deal with the problem much more efficiently.

So perhaps you, or some interested Prime adept, could write a Prime programme to partition a 44,444 digit integer in palindromic parts more efficiently?

You're missing the point. Machine efficiency isn't what I'm after. I need a language that is efficient for me. Having a hard time figuring out what a program does years later isn't efficient for me. HPPL is easy to understand even if you haven't written the program or have forgotten just how it's supposed to operate. If a program runs in 21 seconds rather than 30 seconds but it takes me a half hour to figure out how to change it, then the slower program that I understand right away is the one I'll use. If you can provide a pseudo code equivalent of your program, I'd be glad to try in on the Prime.

Quick addition: I'm not trying to say that RPL is no good or that it shouldn't be available. All I'm saying is that for me HPPL is a more efficient use of my time and a better way to program.

[grsbanks]

(02-18-2018 05:25 PM)acapde Wrote:  BTW, did you try this?

https://www.hpcalc.org/details/8716

It's definitely on my "to-do" list. RPN would make the machine much more bearable

[acapde]

(02-18-2018 09:04 PM)grsbanks Wrote:  

(02-18-2018 05:25 PM)acapde Wrote:  BTW, did you try this?

https://www.hpcalc.org/details/8716

It's definitely on my "to-do" list. RPN would make the machine much more bearable

I know the pain, I've been using RPN since the 80's and can't/won't go back! I now use the 39gs regularly in RPN mode!

[brickviking]

(02-18-2018 05:48 PM)toml_12953 Wrote:  What would you use? That horrible write-only language where everything is in
« » ? RPL is good for one-off calculations and I use it for those. I can see the attraction of building functions like layers of an onion. For permanent programs, though, give me a self-documenting language where I can go back and immediately see how I did something in a program written months or even years ago.

For permanent programs, I'd almost be tempted to agree with you, but once you get good at RPL, "reading" it is probably nearly as fast as reading C or Java, and has the benefit of working on far lower spec machines.

(02-18-2018 05:48 PM)toml_12953 Wrote:  People may say that RPL is closer to the machine and therefore more efficient. I say that nowadays, the hardware is advanced enough to make that a moot point. My time is most important. I want the machine to do the busy work of parsing, hierarchy, etc..

Yes indeed, your time is important, and you must decide best how you'll use it. My problems with RPL aren't with actually reading it, it's more with holding enough of the program in my head so I can figure out what it does, though I tend to have that with most languages. On the 50G, I find that RPL is a good "compromise" for getting power out of the machine for only the negligible tradeoff of having things on a stack by default and actually learning the language.

I tried things on the 39G. I really did. I got so frustrated with its educational slant and aplet-based interface that I simply gave up in disgust and went back to my 9750. Perhaps I didn't give it a long enough trial.

(Post 179)