z88dk forums

I've just played around with uip a bit, and the main problem seems to be IP address manipulation (easy fix is to use a u32_t and chase the propagating changes). I've got most of the kernel compiled now and it comes in at about 17k or so without a device driver and library routines, so say 20k when e...

I'm also thinking of way the LINE directive (in the assembler) can be used to autogenerate the debugger breakpoints so a recompile isn't necessary (which would make debugging a more pleasant experience). I've made a change to the assembler to create symbols when the LINE directive is met, if you co...

The latest version of Zsock is up in CVS here: http://cvs.z88dk.org/cgi-bin/viewvc.cgi/?root=zsock As I left it, it worked on the z88, CP/M (via a hacked zxcc) and Linux. I had it running on a ZX at one point in time as well. I think the kernel itself is probably only useful as a historical curiosit...

A pragma directive would probably be the way to go. Something like: #pragma enable_debugger /* This will tell the crt to link in the debugger */ .. some code (not debuggable) ... #pragma debug_on /* Enable LINE symbol generation and turn off peep hole:ing */ .. Code which we want to debug ... #prag...

Hi dom! I have now commited a few fixes that should make this work. One still must patch "my" test_crt0.asm into lib/ I was a little weary to change this file since it was not "mine". Maybe we could make another target, with a test_debug_crt0.asm? I need help for this though... ...

Hello again! I was on digest so I didn't get this message until now. I looked into the code and it seems you need to compile the test/machine directory first, since we use the .o files in the dbgbin. Maybe a library in test/machine would be a better way? About CMD_DBG I think you need to update tes...

Hi, yup, that's a known issue that's been fixed in CVS. Since you're on a 64 bit platform I would strongly recommend that you take a nightly snapshot since a lot of 64 bit issues have been resolved since the release of 1.9 (rather than fix up this one particular issue).

Sorry for the "mail bombing" I'm doing recently :D Me and Marcello are going to have a little more free time and would like to try and build an updated installer for win32. Is it ok if we base it on the current CVS snapshot ? I have no objections to an installer being made from CVS, but p...

And so I read for the 3rd time, and this time I don't get my z's and s's mixed up. mzx, that makes more sense!

I can see the error now as well, looks like a linker problem so I'll investigate it.

EDIT: Should be fixed now.

No worries, it's easy to do (I do it all the time!). I'll sort out the build system to send out meaningful emails when it fails.

I didn't try the MSX thing - but f_yesno is in the ZX math library which I'm guessing shouldn't be involved in an MSX build?

Yup, rand() is available (look at the source/comments in z88dk/libsrc/stdlib/rand.asm), if you're using the generic fp library then fprand() is also available.

Actually Stefano, I've just noticed (should sort out the reporting I think) that the build has been broken since the 18th - it looks like you didn't check in zx_colour

Andrew, I think the wiki http://z88dk.org/wiki/doku.php?id=library:interrupts has got an example of how to do interrupts with z88dk, it looks like you can do pretty much whatever you like. I was sorting through some old source code a couple of weeks ago and came across my multiple processes on the Z...

I get the same problem with the ZX, but no linking problems with an msx target. I've just tried it out on the cpm target (commenting out the graphics calls) and it does run, but never ends - the conditional check isn't met for FP numbers (make it <). Similarly running on that mode in the ZX the prog...

Yup

I think this is the same as jannone's first problem.

Basically either you need Z80_OZFILES still set (with the trailing /) or add -IDESTDIR/lib to LINKOPTS in the config files and regenerate.

That looks like another 64 bit portability issue. As you might have guessed, I'm slowly working through the issues that pop up (and with such old software there's bound to be a few!) so apologies, hopefully the next release will be 64 bit clean. If you do a cvs update on src/z80asm and then rebuild ...

Almost there, it looks like you need to set Z80_OZFILES to {z88dk}/lib/

I'm guessing it's a 64 bit install?

Two solutions.

1. Compile as a 32 bit app: export CFLAGS=-m32 and rebuild
2. Apply the patch here: http://www.z88dk.org/forum/viewtopic.php?id=3534 which hopefully fixes the 64 bit problems.

I've finally created a 64 bit patch for sccz80, if anyone has a 64 bit machine I'd appreciate some feedback on it. I've compiled the libraries and they create the same result. The patch can be downloaded from: http://nightly.z88dk.org/patches/ To apply: cd z88dk/src/sccz80 patch -p0 < sccz80_64bit.p...

I think the flag -notemp is what you're looking for - stops it using the temp directory.

You may also want -no-cleanup which will stop it deleting the intermediate files.

BTW, giving the -h option to most of the programs (except zcpp and copt) shows all the options for the programs.

Ok, I've just tried on a pristine winxp box which has never had z88dk on it before.

It looks like you still need to set Z80_OZFILES=c:\z88dk\lib\, I'll look into this and try and figure out why the new code doesn't work they way it's meant to.

Ooops, missed that post!

The -\\ thing looks like I was a tad over zealous in replacing the / with \ .

You'll need to replace the -\\ with -// on the OPTIONS line of the z88dk/lib/config/*.cfg files for the targets you're using.

Alternatively, grab a nightly which hasn't got that problem.

Nope, we've just not get an installer this time.

You just need to unzip the zip file to c:\ and setup the path and environment variables.

can I suggest that the "test" target might be a good initial target since it's pretty much a raw z80 and is unencumbered by a current io implementation as well as being shipped with z88dk so it's available for everyone. I can do that but the only trouble with a raw test target is it's dif...

Very nice (ideas and code) I think an example implementation will make things a lot clearer - can I suggest that the "test" target might be a good initial target since it's pretty much a raw z80 and is unencumbered by a current io implementation as well as being shipped with z88dk so it's ...