Welcome to Cemetech.tk, the site for all things calculator!
Discuss today's news on the Cemetech Message Board.
Latest released version of Doors CS (v4.9.0 delta) available You can now create mobileTunes v1.0 TI-83 song files. Click here, create a song, then compile the result.
At long last, I have managed to make some progress on Doors CS for the TI-83+. I discovered what the bug was: all routines worked fine, but as I suspected, there was an underlying problem in what it was doing. It needed to swap the program to higher mem than Doors CS, not lower mem as it needs on the TI-83. I now need to devise a routine to swap it the other way around so that the calculator won't crash if prgmZDCS is in high memory. This should take a relatively short time, and I look forward to a projected release date in the next two weeks. The current beta (beta 3) can be found at the Doors CS Member Page.
Further Doors CS v5.0 TI-83+ Progress
February 28, 2005
Discuss
So now I've tried three separate ways to swap a program from RAM into UserRAM ($9D95), and each works on VTI and the TI Flash Debugger, while crashing on a real TI-83+. I have finally discovered that I have been forgetting to decrement bc (program size) twice to account for the ASM token ($BB,$6D), so I hope that if I fix this and debug why the VAT is being corrupted, I can make it sucessfully not crash on the real calculator. I'll post more when I have tried this and have results (or lack thereof).
In other news, The Cemetech program archives have passed 1,000 downloads since early December, and there are now 61 Cemetech members. My ticalc.org downloads are nearing 375,000.
Since the last news article, I have been exhausting pretty much every possible bug that could be causing the Ion program crash. I thought that the swap routine might be the problem, so I completely rewrote it with bcalls in under 79 bytes, yet it still crashed in an identical fashion. The current evidence:
Ion routines are not causing the crash Crash occurs with all ASM, Ion, and Doors CS ASM programs Therefore the swap routine used in all three of these program types must be at fault There are absolutely no coding errors in the routine - I have checked tens of times Thus there must be wrong with what it is doing, not how it is doing it Crash appears to occur Crash occurs before actual program code is reached Size, to, and from are all set correctly
Here's the steps I'm planning to take as of now to determine the problem and how I'm going to solve it:
Test with programs of different sizes to determine exact crash size It seems there is no exact size, so what is the range? What causes the change Determine exactly where in the loader the crash is occuring Is ROM 1.18 moving instead of copying ZDCS??
More to come.
Doors CS v5.0: At The Verge of Completion
Feburary 8, 2005
Discuss
At some point last week, I believe around Wednesday, I realized that certain Ion (and Doors CS) programs crashed when run on the TI-83+ edition of Doors, yet only on a real TI-83+ running ROM 1.18. When tested in VTI with 1.12 and the TI Flash Debugger with 1.16, nothing of the sort happened. This led me to deduce that the problem could not be related to the archive, since the programs did not crash in the Flash Debugger. I began by trying to brainstorm what might be the problem. I looked through the Ion routines I was using, noticing a bug in the iRandom routine and fixing it. This cleared up a crash problem on the emulated calcs, but the crash on the real calc remained. I began to add modules into Doors CS to debug it. First I had it check and display the mem location from which programs were being copied. This showed that it was working correctly. I repeated the procedure with the program size, thinking perhaps it was creating programs of size 0 or infinity. Once again, the check showed that size was being registered correctly. In the meantime, I had a TI-83+ port of the Password Protection SE and DCS Pong v1.0, ironing out their bugs, and finding the rand error. After trying out a couple of sample programs and seeing that they only crashed when actually calling the library routines, I thought that perhaps the library table was being copied incorrectly. It seemed unlikely since it worked on the emulators, but I first inserted a debug segment in Doors CS that analyzed the lib table. It showed that the pointers were being set correctly. I then checked the first two bytes of each routine as it was pointed to; once again, it was correctly set. In desperation, I put the routine into a sample program, thinking that the swap was causing the crash. Still nothing; the routines gave exactly the correct value. So here I am now, stumped and frustrated.
In other news, TI-Freak8x helped me in the archives project by uploading the screenshots for all of my programs to here from ticalc.org. I have also helped him with a neat project of his, a page in which he is trying to catalogue all of the TI calculator community sites on the Internet. You can check it out right here.
Discuss today's news on the Cemetech Message Board.
More News...
Latest released version of Doors CS (v4.9.0 delta) available 
Ion routines are not causing the crash