Glad to find something that works on 2.03. While we're on that subject... anyone know what rom is required to get Street Fighter 2 working?

streetfighter needs rom 1.00 or 1.05, but it's not worth it. Although it's a great game, it crashes continuously. You need the archive saving feature of rom 2.03 to use streetfighter, or the program will be deleted every five minutes at least. You can try rom version 1.00 with archive utility v. II, but archive utility doesn't always work.

Would someone please do me a favor and answer these questions!!! Does bomberboy work under hardware version 2.00 and AMS version 1.05? Does it work under universal os for hardware 2?

I'm pretty sure it will work, if you read jon's comments on this article, you can try to get it to work for any rom, hardware version.

I've been meaning to mention this for awhile now. I see Grent Jones himself in here interacting with users of his program. It's such a great way for the author to get feedback and for users to get support. It simply enhances the experiences of both sides. But this isn't possible with every program that is released. Maybe it would become too crowded, or take too much disk space, but it would sure be a nice feature to have a discussion area for each file that TICALC archives. They could put a link in that info screen of each program when you click on the little file icon.

I know that this is out of topic, but I don't know where to put this, and it is important. My theory about HW2/AMS2.03 is proven as true from various sources. HW2 calculators really have built-in a hardware device which does not allow that ASM programs longer than 8K may be executed in RAM. This device is turned on somewhere in AMS2.xx (that's why the protection don't work on AMS1.05), and in according to experiments, it seems that it can't be turned off (without removing batteries) after turned on once.

But... it seems that this sad story will have a fortune end. Probably no protection is perfect. The protection device really can't be switched off, but Julien Muchembled (author of Universal OS) points to me that using one nasty trick it can be bypassed!!!
Engineers from TI forgot to implement something, and this fact allows bypassing the protection!

Currently, I KNOW how to made arbitrary long ASM program which works on damned HW2/AMS2.03 combination. You want a proof? I just uploaded on ticalc release 1.1 of my ScottFree driver for adventure games (archived in "advint.zip", TI-89 Assembly games) which
in 14K long and works on HW2/AMS2.03 (it will be available in the archive in a next few days).

As I told, I know how to made arbitrary long asm program by myself, but it is still the problem of executing EXISTING programs without serious modifications of themselves (the programs themselves are not so problematic to modify, but libraries ARE problematic). The problem of making a shell (like UniOS, Doors etc.) which will be STABLE, and which CAN execute EXISTING ASM programs on AMS2.03/HW2 is still open, because the nasty protection device can crash calculator on such terrible way that system recover is possible only by removing batteries (this happens when protection device detects "illegal" behaviour of the program). It seems that making "universal" shell is an amazingly hard task, but not impossible! Note that both Julien and Xavier works intensively on this problem!!!

Could someone please elaborate? I've never heard of this hardware limitation to 8k. I (naively) assumed that the 8k limit was only enforced through software. How is it implemented, this hardware 'lock'? Is it some kind of logic detecting jumps to 'ilegal' addresses, or what?

Just this.There is a logic device built into HW2 calcs.
This device is controled using 64 bits located on 0x700000. Each bit correspond to one 4K chunk, and if this bit is set, PC cannot be there. AMS 2.03 sets these bits on such way that the only possible place for the program is just on the top of the RAM (so, the maximal length of the program is 8K). Note that AMS2.03 always relocate programs at the top of the RAM. The main problem is that it seems that these bits can't be cleared after once set. When this logic device detects that PC is in "forbidden" zone, it masks out nearly whole ROM out of address space, so calc can't read the ROM, until the batteries are removed. Really nasty. I hope that this explanation is enough...

A good question at this point is: Why did TI do this?
I have heard the thing about running flash apps from ram but that is probably not the real reason. Even without the 8k limit, you could not run a 350K flash app from the ram. The only apps that would fit are the language localizers and they are free anyways. Is the 8k problem both a hw and a software lock? On hw1 the programs just get an error, and on hw2 they lock up. This makes it seem like TI really hates their customers, which is odd because their customer support people seem friendly enough.

-Sean Barnes

Don't take this the wrong way... I love calculator stuff and especially ASM games as much as the next guy here. But...

I don't think TI hates their customers. I think they hate customers that reverse-engineer their products and make programs that TI never dreamed of making. And we make them free. Well, TI says, "Hmm... we could make some money off this! Let's rig all our calculators so that those stupid kids can't make big games that we wish we could have thought of first!"

Personally I think TI is pretty stupid to do this. Although they make most of their money off mass-purchases by schools and students, I'm sure many students have chosen a TI over a HP or Casio because they saw their math-genius friend playing mario on his TI.

Oh and yes, as I have found out from multiple companies, especially Microsoft, Customer Support != The People That Make The Product. They should, but oh well.

Don't get too pissed off at me for this,

- Jeff

I think it's not the reason of the 8Kb limit . because you can make a program that uses many little files of 8kb , instead of making a program that uses one BIG file of 40Kb or more . Maybe it was necessary for something , you'll never know ! I bought a TI89 not for Mario or Sonic , but for it's power . Always talking about MATH . HP49G maybe can transform a triginometric function to sin/cos identity , but who said that TI89 can't do this ? try this : tan(x)+tan(y)" . HP49G do nothing with this(you will need a comand) , while TI89... Anyway I like to play games . And maybe TI doesn't want to loose it's Image with a calc that can be used as well as a Game Boy , I think they are against videogames , "the enemy of education" Maybe.....

Yea... I didn't buy my 89 for games either. I had $150 and nothing to do with it... and I was at Staples one day looking at some joysticks they had. I saw the 89 and went "Hmm... what's this?" I had already gotten an interest in programming my 83, and I saw how much this one kicked ass. So I bought it. Later I discovered that there were Mario and Sonic, which IMHO aren't the best games anyway.

- Jeff

Really, did TI learn from history? They had in 1983, desktop computer TI99/4A without PEEK and POKE, so the only way to assembly programming was purchase of hardware module. The machine had fiasco, although it was the cheapest 16-bit computer at this time.

In 1982 they had calculator TI88. Many hackers of their previous model TI59 discovered that TI88 has no hidden capabilities that TI59 had and TI88 stopped with production after only five months!

well as long as everyone here is off topic I might as well be too. I have one of those TI99/4A's but it doesn't work I don't have PEEK and POKE though I have music maker and some other game.
~Matt~

In the words of my friend "if I could shoot good, dribble well, lose 100 pounds, and grow 10 inches I wouldn't be a bad basketball player".

I wonder if you can still execute code which is located in archive memory. I know that it works on my HW1 calculator.

Good question... I guess you could, based on the fact that flashpages can have the function of ROM, App-Space, and Archive, and that the partition between Apps and Archive isn't static.

Unfortunately, you CAN'T. Checked :-(

Actually, on my HW2 AMS 2.03, It worked fine when I ran cblaster() after I archived it. (That is what is meant by 'running code from archive memory'' right?)

Do you think that if we can get a bunch of people to write to TI asking to remove the 8k limit in a future rom version, then they will do that? There reasoning for having the 8k limit ("So that you can't run flashapps from RAM") is a bunch of BS.

that's not it what is meant by "running a program from archive memory".

when you archive a program then run it using doors or uos, the kernel will unarchive the program (or as with the new doors, just make a copy of it to RAM) then run it; he's talking about running something from the archive without copying or unarchiving it

So you mean running from archive like running flashapps? (And as of now, no kernals work on HW2 AMS 2.03)

The fash apllications are programs , but you cannot see the files in it , for example , statistics takes 300kb of archive memory but it is not one long file , it's many 8kb little files that you cannot see , and the application is executed in RAM as well as normal programs .

Not only "when you run it using doors or uos". Even TIOS does this when running the simplest "nostub" programs.

Why archived program can't be executed from the archive memory? The devil device built in HW2 can be controlled by software to determine where PC can be and where can't be, but this device may be controlled only from TIOS (more precise, from trap $B), and any attemptions to gain control over it out of TIOS (e.g. from user RAM) fails. This device can be controlled only from strict address space!!! Also, it seems impossible to jump in the middle of the TIOS routine which controls it. The reason is too complex to be elaborated here...

What about a hack a'la Archive Utility, i.e jumping into something other than code, that happens to do the right thing, and is located right. How narrow is the apropriate address band, by the way?

We are now working just in this direction...

Is there a (technical) reason that the calc has to CRASH when the hardware detects the over 8k program? I mean could TI have made it just give you an error message, or is the only way that that sort of device could work (by crashing/clearing the ram). And is that why there is always that black line across the top when it crashes?

Yes, there is a vicious "techincal" reason: if calc just give you an error message, it can be trapped by software, and the protection can be breaked easily.
But, on following way, there is nothing which can be done :-(

Well, while we're on the subject of breaking limits, i have a related question: Did anyone try to break the 384K archive limit on HW1 calcs?

That limit, while it might be a little annoying, is definatly not as importaint as the 8k HW2 limit, for a few reasons:
1) There is no reason TI would want that limit, TI is probably working on that now anyway, I don't think there is a real reason for it (just a bug)
2) If you have the stats app, then that is not a problem (it is very large)
3) You can at least play some games / run some programs
4) 384k is not really restrictive (at least compared to the other calcs)

No offence, but I think that we should only worry about breaking one limit at a time.

I really can't think that TI would want that limit, but I must point that this 384K limit is also mysterious. AMS2.03 checks whether the calc is HW2 calc, and if it is not, it does not allow archiving below certain address. But, it seems that the program CAN be archived below this address if this check is by-passed. So, I want to say that it seems that this is a software limit. Again, I really can't think about any reason that TI would want that limit, but it seems that they also implemented this limit by themselves??? But, this is a hyphotesis: maybe I am wrong. Or, maybe TI want to "prove" that HW2 calcs are better than HW1, with a goal to force buying HW2 calcs which have built-in various devil protection mechanisms...

Sounds quite resonable to me. And i've got the feeling that it's also connected to the issue (that we all know and love ;-) ) of flash app piracy. Since hw1 calcs haven't got the 'proper' restrictions for running code, it would be quite easy to put a pirated EEPro, for instance, in archive mem and get it to run through some tinkering. By restrikting the sise of archive mem this becomes more complicated. (Though not impossible, of cause.)