KnightOS Updates

DarkestEx · March 30, 2015, 02:30:19 PM

So we can try implementing parts of posix in the libc, right?
I mean we could need stdin, stdout, stderr and arguments, right?

SirCmpwn · March 30, 2015, 03:04:43 PM

Quote from: DarkestEx on March 30, 2015, 02:30:19 PM
So we can try implementing parts of posix in the libc, right?
I mean we could need stdin, stdout, stderr and arguments, right?

There's no formal support for arguments, but there's some proposals. There is no stdin, stdout, or stderr. The file API is currently hardcoded on KFS operations and hasn't been abstracted out to allow for things like that yet.

You are welcome to help implement parts of POSIX in libc, though. Keep in mind that C support is still experimental in KnightOS and there are several important problems with it, and you will run into frustration if you try to embark on serious C projects on KnightOS.

SirCmpwn · April 22, 2015, 03:59:28 AM

C support continues, and is starting to become something usable for actual projects. More sophisticated programs are working now:

Bonus: I got it to work for the most part on TIOS, and I plan on adding support for writing programs in C for TIOS formally. Also, since KnightOS's libc is statically linked, you can use (most of) the POSIX functions provided by KnightOS on TIOS.

Snektron · April 22, 2015, 10:21:48 AM

Ooh nice

Duke "Tape" Eiyeron · April 22, 2015, 11:34:21 AM

Mmh, puts is supposed to insert a newline after the string. :-°

SirCmpwn · April 22, 2015, 02:21:49 PM

Quote from: Duke "Tape" Eiyeron on April 22, 2015, 11:34:21 AM
Mmh, puts is supposed to insert a newline after the string. :-°

Yeah, I didn't bother looking into why it didn't work, that was just a proof of concept.

Duke "Tape" Eiyeron · April 22, 2015, 04:55:53 PM

Yeah and it looks very good. I can't wait to have some test runs on this sdk.

SirCmpwn · May 07, 2015, 08:12:30 PM

Minor updates:

Ivoah contributed a change to the castle that prevents it from showing the "you might lose work" warning on shut down if there are no running threads other than the castle.

libc is progressing. I've added support for SDCC built-ins from string.h (like memcpy), as well as adding a little bit more to ctype.h. C support in general is improving steadily, with bug fixes in scas and kcc going on. Handling for include paths for C programs has also changed a bit in the SDK to be more flexible and less stupid.

Also, the TI-73 explorer was tested with KnightOS. I expected it to work, and it did, so that's good. No updates on the 84+ CE front - I still don't own one and there are still no public exploits for installing unsigned operating systems.

Dream of Omnimaga · May 07, 2015, 09:30:54 PM

Glad to hear about those updates. As for CE support I hope it's not like the Nspire and doesn't require nLaunchy.

CKH4 · May 07, 2015, 09:53:19 PM

I may be able to test knight os when you get the ce version going. I don't have one yet but I'm planning on getting one soon.

SirCmpwn · July 19, 2015, 12:26:10 AM

Kernel 0.6.8 released

This kernel adds mature date/time support and improves filesystem support.

Features

New syscalls: isLeapYear, monthLength
New kernel panic screen (screenshot)
convertTimeToTicks is no longer a stub

Bugs fixed

Several bugs in 0.6.7's date/time support
reassignMemory checks that the target thread exists and returns an error if not

Deprecated

`findFileEntry` and `findDirectoryEntry` are deprecated and now behave as no-ops. Internally, the kernel now uses `findNode`, and KFS internals are no longer exposed to userspace.

Download

kernel-headers-0.6.8.pkg
kernel-TI73.rom
kernel-TI83p.rom
kernel-TI83pSE.rom
kernel-TI84p.rom
kernel-TI84pSE.rom
kernel-TI84pCSE.rom

Kernel 0.6.8 on Github

Disclaimer

This release, like all other kernel releases, is only relevant to developers. The kernel also will not run on its own, so if you put this in an emulator you won't get much out of it.

Dream of Omnimaga · July 19, 2015, 06:05:19 PM

Good to see updates.

Do you know approximately how much percentage of the project is done so far? (by approximate I mean something like 20-30% not a specific percentage

)

SirCmpwn · July 19, 2015, 06:06:30 PM

It's a bit weird to ask for a percentage. I don't even know how I'd begin to calculate that. I don't even know what "done" means.

Dream of Omnimaga · July 19, 2015, 06:07:56 PM

Oh I mean until it's ready for public use in class and stuff, like Axe Parser 0.2.0, for example (the one that made ticalc.org front page), unless it already is? I didn't install it yet, since I needed to run TI-BASIC programs.

SirCmpwn · July 19, 2015, 06:14:21 PM

Depending on your use-case, it's ready now. It runs on all of my calcs and the same can be said for several other people. The important milestone is probably math, and that depends on several other things being in place. Kernel 1.0.0 will be the biggest milestone the project has seen so far, and that requires USB support, I/O networking, fully fleshed out filesystem support, better compression support, a mature C API, and some major refactoring projects (refactoring internal kernel data structures, for the most part). For math, we first need to flesh out the package manager and add support for transferring files to calculators post-install, as well as bringing some kind of BCD support in the form of kernel support or a userspace library (or both). corelib should also see more development before we embark on that.

Before I'm willing to call it "1.0", we also need kpy support, a settings manager (in progress now), fileman needs to be fleshed out more, the castle needs to be finished (it's nearly done, but depends on kernel 0.7.0 features), and bed needs to be finished (it's not even close to done), along with all of the things mentioned in the last paragraph.