In 1996, with the arrival of the TI-83 and its string support, two new bugs were introduced. First of all, there was the well-known Equ>Str bug, causing large files and crashes to occur after pressing Y=, CLEAR, GRAPH, Y= again, typing zero in Y1, typing Equ>String(Y1,Str1 on the homescreen then ENTER 4-6 times.

The 1996 TI-83 also introduced another string-related bug, occuring when recalling 2-bytes tokens (such as lowercase letters) with only 2 bytes of RAM left. During the recalling process, after the first quote appears, leaving 1 bytes of RAM left, the calculator would attempt to recall the next character, regardless of its size, even if only 1 byte of RAM was left. When it was a 2 bytes token, the result was this:


While the Equ>Str bug was fixed in 2002 on TI-83 Plus OS 1.14, the 2-bytes token RCL glitch still remains. Not only that, but the latter bug is still present on test versions of the TI-84 Plus C Silver Edition. Basically, it has existed since before some Omnimaga members were born.

In fact, the arrival of the larger screen and colors even introduced more bugs. For example, the menu() command now supports 9 choices instead of 7, but the last two display as "??" and cause an ERR:LABEL. Also, as reported before, if a Pt-On() command uses colors, those colors will show as black until the program is stopped completely. Another issue is ALPHA scrolling scrolling by 7 lines instead of 9 in the program editor and how setting 0?Ymin:1??Y incorrectly sets Ymax to 165 instead of 164.

Thankfully, the calculator sample that KermMartian, Critor and Phero got might not necessarily be the final product, so if they reported those bugs to TI, it is possible that they might fix them before it hits the shelves. But knowing TI, there is also the possibility that they will take a long while before fixing them.

There are good news, though: The following math bug, for example, was fixed on the TI-84 Plus C Silver Edition:
It was, however, reported back in 2008 by Critor. Will TI adress remaining issues (including older ones) faster, if not done already?

Also, it has recently been reported by Vijfhoek on OmnomIRC that the sample of the TI-84+CSE that he had the chance to test at school showed no OS version in the About menu and the calculator in overall seemed slower than Phero, Critor and Kerm's, meaning that it either used an older OS and/or different hardware.

It would be dumb of them not to fix the bugs before they release.

Fixing some bugs is quite a long process for TI (and any other big company), especially if the fix is not trivial.
If TI is already ready for mass productions, well, the ones in the shelves won't possibly come with bugfixes.

Maybe in an early (meaning, not so long after the release) OS update (like "4.0.1") ?

Maybe the samples critor, Kerm, etc. received were also "old" ones, and not the latest with nightly OS build, with bug fixes already....

Anyway I believe KermM has reported (or will soon) them.

Yeah, this is why I tried the 2-bytes token bug on my TI-85 as well and then tried it with Y vars on my 82. However on the 85 those tokens were finally 1 bytes so the glitch couldn't be triggered, and on my 82 it seemed the bug didn't happen when recalling the content of Y-Vars.

Nope, he's a student.

Awesome! Are you a teacher adriweb?

As Lionel said, I'm a student (First year of engineering, I finally have some CS at school )

Quote from: adriweb on March 05, 2013, 05:47:50 am

Fixing some bugs is quite a long process for TI (and any other big company), especially if the fix is not trivial.
If TI is already ready for mass productions, well, the ones in the shelves won't possibly come with bugfixes.

Maybe in an early (meaning, not so long after the release) OS update (like "4.0.1") ?

Maybe the samples critor, Kerm, etc. received were also "old" ones, and not the latest with nightly OS build, with bug fixes already....

Anyway I believe KermM has reported (or will soon) them.

Well it depends of what bug it is. Math ones might be harder, but for the 2-byte token glitch, it's just a matter of checking when recalling a string if there's 2 bytes of RAM left for every character. Since recalling Y-vars causes no problem, I bet they could just copy the routine from that for strings (although strings recalling includes quotes).

Yeah, I guess those could be easily fixed