What you are asking isn't exactly modest, but I do see your point. However, on the point of functions being able to return multiple values is one that probably has been adressed before. Unfortunately, I doubt this will change. Since most computers and design paradigms are allready one result per function (thus the whole basis of what a 'function' is), it would be a pain in the butt to write a new compiler than would handle this multiple return value system. (At least for TI), however, if TI were to write functions that returned lists of stuff instead of a value, or a matrix instead of a list, that would achieve the same result, wouldn't it?

Once again, this was kind of a rant written in a hurry (and I didn't give it a title)

From our own functions, we can return lists, but TI's do not. I think that Matlab's design with regard to this is nice. The function outputs a vector of values (and is, thus, a true function) and the user can use as many or as few as he wants, storing them to separate values.

I tried this on the TI-89 a while ago using vectors and lists. A function which returns a list can be executed in a program as
foo(bar)->a
but not
foo(bar)->{a,b}

While the problem may be worked around by adding a few lines of code, if the function written is intended for user use (and not to be called by another function), it's really inconvenient.

Also, if the elements of the list or vector of outputs are themselves anything but expressions, strings, or numerical values, this method simply will not work. If I wanted to, for example, return matrices O (matrix when multiplied by the original yields U)and U (reduced row echelon format)from Gaussian reduction, I could not return a list {O,U}. When I try to enter such a list from the home screen, the calculator returns an "invalid list or matrix" error, and the same would occur in a program. The "data" type variable is a step in the right direction, but current data items cannot be lists, vectors, or matrices, and are not easily accessed from the home screen or even from within programs. The "data" type seems no better than using a matrix, in this respect--it's two dimensional and the elements can only be expressions or numbers.

Again, I realize that there are reasons for these limitations (having to do with the calculator's ability to allocate memory to variables as the user creates them?)

I thought that using indirection for output, C-style, would work, but it gives trouble, both in TI's built-in functions and our own. For example, the built-in QR and LU decompositions cannot be run from within a function. Using the command 1->#a, with a being a string and an input argument, also gives the error "Invalid in a function or current expression." I presume that there was a reason for this, but I wouldn't know.

-Ben Kalafut

> I thought that using indirection for output, C-style, would work, but it gives trouble, both in TI's built-in functions and our own. For example, the built-in QR and LU decompositions cannot be run from within a function. Using the command 1->#a, with a being a string and an input argument, also gives the error "Invalid in a function or current expression." I presume that there was a reason for this, but I wouldn't know.

This is because functions are too limited. I would have gotten really far with my strlib (a relatively neat but inevitably pointless project to do arithmetic on strings, thus of infinite precision) had functions been able to use expr() or even string(). C/C++ got this right -- don't limit stuff.

But why this limiting of functions? Functions can be used in, say, solving an equation, and you wouldn't want your function to modify a variable that was being used, so TI thinks. But the fact of the matter is that usually you wind up writing the function yourself, and you may _want_ to change some global variables while still returning a value. Maybe it's because AMS was designed so that stupid users wouldn't cause themselves too much trouble and go whining to TI for help. But we're not stupid (I hope) and a bit more power in functions, and overall, would be appreciated, maybe as an "Advanced" mode.

Kenneth

Actually, the title "A Modest Proposal" was a reference to Jonathan Swift's satirical essay in the 1800's about overpopulation in London.

He offered an outlandish solution (boiling kids alive and eating them for meals) to show that solving the problem was no small task.

"Solving" this "problem" is no small task. :)

--BlueCalx

Just buy a mini-laptop. They are getting awfully cheap these days. I don't think anyone realizes the main reason that this device, or any other handheld (ie. Gameboy) uses a really fast processor. Do you actually think a laptop could come anywhere near running for a month on 4 AAA batteries?

What kind of Gameboy do you have? The newest one (G.B. Color) is faster, COLOR, and runs off of 2 AA batteries for longer than 1 month (usually 2). Plus, buying a "mini-laptop" (if one exists in the first place) would be a huge waste of money to complete simple oporations that can be done on a calc. Also, i foung that programming for a calc (82 and 83 for me) is a HELL of a lot EASIER than trying to program things on computers (desktops/laptops)...

You probably don't have a Gameboy Color, cause anyone with one knows that NO batteries will last a month on it, even with conservative use. I've found that a set of 2 Duracells gives about 10-15 hours of powered-on usage...and remember that a calculator is often left on, waiting for the next command, further sucking batteries. No, I think color is secondary to screen resolution, which is not as much of a battery drain as a color LCD.

How much power a Game Boy game takes up is largely dependent upon the game itself. You could write a game that would run for at least 30 hours if you wanted to, but not if it used all the CPU time available and had features like sound.

The functionality I think the 89 lacks is a true abilaity to deal with infinity. Have you ever tried to perform symbolic integratation with infinity as a bound? For almost all functions, the calculator doesn't get it right. It handles numerical integration well.
Hey, I should point out that I was able to solve a system of equations on my 89 that MATHEMATICA couldn't do (at least using their standard array of Solve, etc functions). I was impressed by that.
Also, of note is that there are already Laplace and Fourier programs out there. They run in basic, so they're slow, but they work! Quite well, I may add.

TI's calculators are ALL about the games. They should stick a 20 mhz chip with 2 megs of ram and assembly support into an 89. Then those games would fly. Does anybody else also relize that TI-BASIC is useless?

You know you're asking to be flamed, don't you?

Maybe, but I think TI probably had a reason not to put in 20 mHz chips in. I don't know what it is, though. :-)

A 20 MHz chip would eat batteries faster than you could shove them into your calculator!

Sounds like a good reason to me. :-)

TI-BASIC is NOT useless. It's for math purposes, not for gaming, and with symbolic manipulation, it's actually easier to use than other programming languages, because you don't have to "dive into" anything. If you want to program good and fast games, use C or Assembler, not Basic.

As far as adding functions to new rom versions, TI did that, but it's kind of lame. If you look carefully at the functions in ams1 and compare them to ams2, you will discover that ams2 adds a function called deltaList (delta is a triangle). I noticed this because I was wondering why it wasn't in ams1 in the first place, since ti ti83 and 86 have it. Still, it would be nice if they added some more functions like you mentioned.
As for the lack of quality math programs, I totally agree with you on that. A lot of them do the same thing as a built-in function only with extra 2k of wasted space. Seriously, if you don't know how to use the solve() command in the 89, do you even know that it has a CAS or what? IMO, quality programs/functions should be space-efficient and original, not replacements for TIOS functions.

Some missing features that i want to see would be square root to a power.
3*SQRT(9)
in example. Yes, i know you can raise a number to a fraction
9^(1/3)
to get the same result,but this approach does not take advantage of the pretty print features of the TI-89.
I think pretty much every lower ti has it(ti 85,86 etc) but the 89 doesnt.

RandInt is also a very nice funtion as well. Almost mandatory in certain games. Rand() is ok, but you cannot specify a upper bound or trials(You could use a for loop, but...) It is included in TI Statistics/List editor, but not for almost half your flash memory! And again, this feature is seen on 83s, 86s.

There was (at least)one other thing missing, but I forgot what it was.

Can anyone else think of anything missing from the TI-89, feature/catalog wise?

P.S. Scroling is too slow too. TG for fastkbd.

A different CPU would definitely be desirable. How about a Transmeta Crusoe? (Sure, it's expensive, but for a top notch calc it rulez).

As for the compiler, stick with TI-GCC [{(don't even think about it)}], and improve it. TI has proven their inability to release such things (The TI-89 Flash SDK comes to mind).

That's a bad idea; TI can just save more money and focus on the m68k processor, on how to make their OS's smaller and better for them; in the long run, when they save money, we (usually; capitalism is never a sure thing) save money as well.

I'll again note to everybody HP has been working with that Saturn chip at 4mhz for quite awhile (I think I saw 14 years in a post somewhere around here), and the HP49g (still at 4mhz) can very well compete head to head with an 89 at 10-12 mhz (whatever it is supposed to be). It also actually does real 3D graphing. Thats something the 89 cannot say. The 49G might be slower in some parts, but it is also faster in many other parts (like the 3D graphing).

Isn't a Crusoe processor a PC class device? It would be nice if it would work, but it wouldn't fit and use up the batteries in a split second. Those things sometimes need ~20-40 amps to operate. Sorry, but the batteries wouldn't last.

While everyone here is talking about really advanced math that I don't understand, I just wanted to gripe about some stuff on the TI-83 that has been REALLY bothering me. And yes, I have a TI-89 to do all this, but I can't always have it for a quiz.

The TI-83 needs an environment where it recognizes variables as unknowns, and not just shortcuts to already-defined values. If I give my 89 something (real basic) to simplify, like 2x/6, I get x/3. But if I put that into the 83, I get some funky number, depending on what I last stored to x. Usually I end up with 0. That so very much sucks.

And while I'm at it, how about a REAL equation solver like the 86 and 89 have for the 83! None of that "has to equal 0" stuff! Grrr...

And Jeff Chai: respond to this and tell me what I missed because I know there's more... I just forget.

- Jeff

If you want x/3 on the ti83, it's not gonna happen, because you would need a CAS, like the one on the ti89. However, the problem with the 83 is that if you reference an undefined variable, it automatically gets set to 0, which can cause an invalid answer. At least they (TI) could have made it like the 86, which recognizes if the variable is undefined.

If they put a CAS in the lower-level graphing calculators, you'd be forced to use a scientific calculator on the ACTs. (Then again, that's what I used, and I did pretty well...) (Then again, the only reason they allow the graphing calculators on the SATs is that they don't want to make the graphing calculator owners buy a new calculator.)

There's a pretty quick way around the has to equal zero thing. Subtract whatever you want it to equal from the expression.

For example, if you want to find the intersection of the polynomial x^4-3x^3-6 and the exponential function, you'd use the numeric solver on the expression x^4-3x^3-6-e^x

-Ben Kalafut

<Sarcasm>

But that requires actual work! I'm a lazy TI-89 bum who is used to just plugging in the problem and out comes the answer. :)

</Sarcasm>

- Jeff

If you want to solve polynomials on the TI-83, download my Polynomial Solver! =) It's located at http://www.ticalc.org/ archives/ files/ fileinfo/ 115/11538.html (spaces included b/c the URL was longer than 40 chars >=) just get rid of them). I wouldn't trust it too much, though. It is in beta, and it does screw up sometimes. But it's pretty fast, for TI-Basic at least... If any of you find a bug, it would be cool to mail me a bug report =)

Hmm... I'll try it, maybe it'll make the math quizzes with the no-89 rule easier. Stupid teachers... why can't I use my 89!

- Jeff

'cause you're a rich, crazy, fool. Just like me. I say we kill the teachers and take over!

Total anarchy!!!

Yes Jeff, there is more. The TI-83/+'s catalog function should contain every math command the TI-83/+ is capable of. Believe me, it can do alot more. Next, the Z80 processor in it is running at 6 Mhz. But the Z80 can go much faster than that. Other Z80's in similar electronic devices run at 10, 15, even 20 Mhz! Military class Z80's run at an incredible 30Mhz! So if TI can lighten up and push the Z80 to the limits...

Long live the TI-89!

Ive heard of a fabled 40Mhz Z80 CPU...

yeah, and 4 AAA batteries aren't going to help you with 30-40 mhz z80. Plus, wouldn't it overheat?

Stick with coding better instead of getting spoiled by faster processors

If your not in advanced math a 89 is no more usefull than a gameboy, learn to do real work with out a calc before you get screwed later when you cant add and subtract, heck i wasent even allowed to use calc's on tests until trig, and even then only on some of it.