Ndless 2.0 for TI-Nspire Clickpad/Touchpad

[ExtendeD]

Excellent, thanks bsl!

The program "idcmigrate" I have made to look automatically for the symbols of a new OS image is based on heuristics and sometimes takes a symbol for another. The address of realloc was wrong.

CAS support for OS v2.0.1 is fixed, all the tests now pass.

Levak: strange. Are you testing on real HW or on nspire_emu?

[Levak]

Levak: strange. Are you testing on real HW or on nspire_emu?

HW :
TI-Nspire CAS ClickPad
OS 2.0.1.60
Boot2 : 1.4

Tested connected from computer, and disconnected from the computer.
I've tried at least my "workaround" with USB 3.0, it doesn't work better.

[ExtendeD]

How many bars of pixels can you count on the screen just before the reboot, and how many dotted lines at the bottom of the screen?

[Levak]

Here it is : an high framed shoot.

EDIT : OH MY GOD... stupid bug : The folder contains a dot. When I removed it, Ndless installs perfectly \o/

[bsl]

The missing call isalnum should be:

cas2.0: 0x102f30e8   
ncas2.0: 0x102f2818

I didnt test it ,but looks convincing ....

[Goplat]

That's isalpha actually. TI's implementation of these functions is basically

Code: Select All | Copy To Clipboard

1 2 3

int isalnum(int c) { return isdigit(c) || isalpha(c); } int isalpha(int c) { return islower(c) || isupper(c); }

so it's very easy to confuse them if you're looking at a disassembly without symbols.

It's common practice for the standard C library to be implemented with one function per file, so the linker can omit any unused functions. That's probably what happened - whatever code used to use isalnum back in 1.7, got changed in 2.0.

[bsl]

Yes I just caught it , just now - sometimes they change register usage , making it harder to track.

[ExtendeD]

Does anybody confirm the reboot Levak gets on real CAS hardware (i.e. it doesn't depend directly on the calculator state)?

[ExtendeD]

Levak, according to your screenshot, the issue happens somewhere in stage1.c, and most probably in fopen since you don't have the third bar created by the fread. Either there is not enough memory to (f)open ndless_resources, or it simply cannot be found.

Could you please try the installation with much less files on the calculator, if it's not a problem for you to delete them?

[Levak]

Levak, according to your screenshot, the issue happens somewhere in stage1.c, and most probably in fopen since you don't have the third bar created by the fread. Either there is not enough memory to (f)open ndless_resources, or it simply cannot be found.

Could you please try the installation with much less files on the calculator, if it's not a problem for you to delete them?

I've said, by editing my post, that I managed to intall Ndless 2.0 on Hardware by removing the dot contained in the folder name.
=)

[ExtendeD]

ut_currentdocdir_addr points to ndless_installer's folder... but with special characters removed
I have no other choice than to add a note to the ReadMe file.

[bsl]

I get exactly the same screen on the emulator when I create a folder
named Ndless2.0 and put the Ndless files in there to run.

[ExtendeD]

Goplat: ok, I see. I hope this won't happen to often with the next OS upgrades (if we ever get Ndless to work on them...).

isalnum is now part of the default static library libndls instead of being a syscall.

[Tuxpp]

I tried the alpha version on my CAS Touchpad (OS 2.0.1, Boot2 1.4), and i got some issues :
  - when trying to install ndless, the installation often doesn't work : the calculator reboot most of the time.
  - All programs started, so no problem for that.
  - In all programs i tried, the touchpad  and a lot of keys didn't work at all.

[ExtendeD]

Thank you Tuxpp for the report.

1) Make sure to disconnect the calculator from the computer before running ndless_installer. What is your average success rate with this condition? It should be around 2/3.
3) Yes, the key remapping is the last step before a full support of the Touchpad.