Replicated. The hook chaining needs to be more careful about register validity, because even if another app's hook says to do the default behavior, it might have trashed register(s). I'll improve the hook chainer further to handle this case.

Great! There is also a bug with Transfrm. That post has two videos to watch. Please tell me if you could replicate the bug shown in the second video with the given instructions for after steps 1-4.

I will edit this post later with bug 3. I have only found three bugs total so far.

Hi Kerm, just interested to know how the performance of your grapher compares to Prizm as of now - I saw the video comparison from almost a year ago - is it still below Prizm's for TI-84+CSE? Is it possible to estimate how it will be for TI-84+CE please?

amazonka: Kerm needs the signing key to make a 3D graphing app for the TI-84+CE, which could take a very long time to get from TI.

Kerm: I am putting all of the Graph3DC bug videos into a playlist so they are easy to access and stuff. Hypothetically, if you were able to port Graph3DC to the TI-84+CE and had the signing key, how much faster would it be?

I believe Kerm had once said the CE was about 50% faster than the CSE, so I'm guessing this would apply for this app as well. It is currently (and probably always will be) slower than the Prizm simply because of the massive difference in the CPUs, the CSE clocks in at 15MHz, and the Prizm at over 50MHz, and can be overclocked to over 90MHz, so it isnt much of a surprise that the Prizm is considerably faster.

Why is the Prizm so much faster? Does TI have any calculators that can compare with that speed, and if not, why do so many people have TI calculators, not Casios?

Kerm: I will do a bug report on the third bug I found soon, probably this weekend. I have been very busy with almost no time to check Cemetech. It will be an edit of this post.

Because of both raw clock speed and processor architecture:
* the Prizm's SH3/4 processor is running at 50+ MHz (overclockable to 90+ MHz), as stated by mr womp womp, whiile the TI-eZ80 series uses an eZ80 processor at 48 MHz but effectively running at 12 MHz or 6 MHz due to waitstates;
* the SH family is of a newer generation than the nearly 40-year-old Z80, which the eZ80 is a mildly small evolution of: the eZ80 definitely adds a number of interesting features, and reduces cycle timings, but the eZ80 remains an accumulator-based, register-starved 8-bit processor.


Of course: the whole Nspire series largely beats the Prizm in terms of hardware characteristics, and the HP Prime beats the Nspire CX, except for the amount of RAM.
The Nspire's BASIC doesn't have graphical output instructions, but the Nspire's Lua does, and despite several known speed issues before OS 3.6, it's not plagued by latency issues comparable to those exhibited by the Prizm's BASIC.


Because both Casio and HP did a number of blunders in development (or lack thereof) which enabled TI to gain market share at their expense. What's more, TI heavily invests on teacher training.

The Prizm would have deserved greater popularity among developers, but it was too weak against the Nspire series (the CX came out several months after the Prizm, it was already under development) and possibly too expensive against the highly popular monochrome TI-Z80 series.
TI waited two years before adding a color model to the TI-Z80 series, and two more years for using a more proper processor to drive that color screen, because they could afford, there was no significant threat to their business model.
Making a dent into TI's market share would require competitors to do far better than the Prizm, Prime or the horrible fx-CP400 (barely an evolution of a decade-old fx-CP300/330, extremely expensive, fantastic slowness in some high school level problems which came out in standardized tests, etc.)


But we're partially straying off-topic here.

It's hard to state the difference in performance between the TI-84+CSE and the TI-84+CE in a single number, because code in Flash runs much slower than code in RAM, writing to the LCD is proportionally much faster than it used to be compared with simple math, and the new processor is pipelined. It's safe to approximate this as a 50% speedup for programs that have both heavy LCD usage and heavy computation loads. TI-BASIC programs that just do math will only be about 15% faster, and programs that are almost entirely graphics-based with very little program logic will be more like 200% to 300% faster.

I found a bug. When entering an equation and then press [GRAPH] it calculates the X,Y,Z-values. When then pressing ON, you will get this:


And another one: when first display a graph, then press [TRACE] and going back to [Y=], the first character disappears or becomes another character.

Thanks! I do indeed plan to handle [ON]->Goto correctly by finding a way to return to the Y= menu.

Nice catch. I'm sure this is related to my use of the cursor hook for tracing, because I notice that before the cursor flashes once, the correct character is there, then it disappears after the first flash.

A new one: when changing from LabelOff to LabelOn, a black line leaves


Also at AxesOn -> AxesOff and BoundsOn -> BoundsOff

EDIT: When in horizontal mode, and then at the graphscreen where black is your background; when you go then to the somescreen by typing "1" or so, there appears some black squares.


EDIT2: when you at the Zoom-screen and then press [TRACE], it don't work and the text "Graph3DC" disappears.

EDIT3: when running G3DC from DoorsCSE, you must ALWAYS install the app, even if you already did that.

And going further...
If you are at the graphscreen, then go to Apps and select DoorsCSE or G3DC, it doesn't opens it and return to the homescreen.

EDIT5: when you are entering an equation, then [ALPHA] [F1]-[F5] doesn't work.

Thank you for the thorough testing! Here's what I have so far:

• Chaining menu (and other) hooks: should save registers around chained hook
  
• When switching from 3D to 2D mode, need to figure out how to mark graph dirty in such a way that graph above Y= menu will be correct.
  
• [ON] -> 2:Goto while G3DC is computing (X, Y, Z) points currently causes junk to appear on homescreen. Redirect to Y= as planned.
  
• Adjust threshold for erasing lines rather than clearing graphscreen
  
• Implement Table mode?
  
• Extra column of pixels with left-margin inverted text in Format menu
  
• When in horizontal mode, and then at the graphscreen where black is your background; when you go then to the somescreen by typing "1" or so, black squares appear
  
• [TRACE] from the Zoom menu doesn't work (and the "Graph3DC" title text disappears)
  
• If you are at the graphscreen, then go to Apps and select DoorsCSE or G3DC, it doesn't open it and instead returns to the homescreen (check interactions with AppChangeHook?).

Not really a bug. Doors CSE juggles hooks while it is running, so Graph3DC can't tell that its hooks are already installed (but temporarily removed by Doors CSE). The only fix I can think of would be a change to Doors CSE, to make it restore all hooks to normal before running external Apps. And this would defeat the purpose of running Apps from Doors CSE, because Doors CSE uses an AppChangeHook to regain control after running Apps.
Not a bug. The OS does not provide a way to enable those menus, for the same reason the equations are not in MathPrint mode.

Can I beta test this program? I have been waiting for a 3D graphing app and i really want to test this one out! (I have a TI84+CSE)

Thanks for reminding me that this project exists! I currently have a number of reported bugs that I need to address, after which I'll likely release a public beta. If I decide to do another phase of private beta-testing, though, you can certainly be on the list (as can solarsoftware, who has nagged me about it for a few months). I must say that learning that it was extremely unlikely we'd be able to release our own Flash Apps for the new TI-84 Plus CE strongly discouraged me from bothering to continue this project, though.

Phst, no... What are you talking about?

It's not even like it's surprising news, though...

No, but I was hoping that this project would help convince TI that it was a good idea for us to be able to make CE apps, since we can add educational value to their devices, not just games. That's why I bothered spending so much time on this project this summer.

Well, even TI promotes calculator game programming, actually. (In TI-Basic, however, yes.)
So, we don't really know what they think about all this. Nothing app-signing-related, though AFAWK...

Thank You, I would also gladly beta-test any programs and/or apps that you are working on/ will work on in the future

Has any more progress been made on this, or is it finished already?