Advogato blog for chaos

18 May 2001

Fri, 18 May 2001 20:06:52 GMT

In order to keep working on make the old quake binaries working, I have to compile some software with libc 5 (Mesa, svgalib, glide, ...) as the old quake binaries were compiled with libc5.

Even if there is some libc5 binaries around for those packages some are old (Mesa and Glide) and do not offer the features of today versions. As I stated earlier, Glide did not work with athlon, but the new version does.

So I repackaged George B. Moody, george@mit.edu, gcc5 package, adding a few corrections and put it here.

With it anyone can do a:

$ CC=gcc5 make

and compile an package with libc5 support out of the box.

It did worked with Glide_V2 and should work with anything that supports libc 5 and gcc-2.7.2.1.

I feel that I have to release it in freshmeat as many other people might be looking for something like this and not finding it.

17 May 2001

Thu, 17 May 2001 15:15:41 GMT

ok. There have being a long time since my last post. I hope to do it more frequently now.

BTW, if you are wandering the Z8 software uart (RS-232) with the lcd terminal worked. It was fun to attach it to external modem and start some conections.

Now I am collecting linux quake stuff and packing it (RPM). The big challenge was to find libc 5 stuff that work with it.

I have already working: svga quake, x11 quake, gl quake, tty (textmode) quake and wm quake (a window maker dock that runs quake!).

The biggest chanllenge is to make glquake work. Since it needs a Glide (VG, VRush, V2) compiled with libc5 and all I could find was old ones (linux.3dfx.com is down, remember :\ ). It works, but not on AMD Athlons without turning the CPU cache off. The latest glides have a fix for it, but I could not find a libc 5 compiled for it.

I am looking for a way to generate a libc5 compiller environment to compile it myself. So it could be improvements.

I am posting all of that at http://www.swi.com.br/~chaos/files right now.

So if you enjoy quake and friends, let me know.

The files are RPM packaged ones tested on Conectiva Linux. It should work with Suse, RedHat, Mandrake, Caldera without problems.

If you use slackware, Debian, or want to extract the files, use:

$ rpm2cpio rpmfile.rpm | cpio -i -d --no-absolute-filenames

This will extract all files into your current dir.

So, right now I am emphasis old quake games, then I plan to move to opensourced ones (SDL quake, quakeforge) and then move to more recent ones like quake2, quake3 and other friends (doom 1 2, wolf3d, heretic and everything that works on linux and it is free redistributable).

25 Jun 2000

Sun, 25 Jun 2000 15:00:29 GMT

I am back after sometime.

As always time is not by my side.
I have to finish a project to Tuesday and it will require a _lot_ of work.

I am writing a multiuse RS-232 device.
It uses a Zilogs Z8 microcontroller, a LCD display (240x128) and (ugh!) a numeric keypad.

I am facing some really unpleasants problems:
- The Z8 I am using do not have a UART. I have no choice about this, so I had to write a software UART to do the job.
The problem is that a _software UART_ eats all available resources and blocks during the transmition/reception.

- I can t use a more complex key[pad/board] without an extra microcontroler to continuous scan its lines. ( an standart IBM-PC uses a intels 8048 microcontroller for that!).

- The tool provided from Zilogs (ZDS) _really_ sucks. It does not work (debugging), without a hardware emulator attached to the serial port.
Well, at least it works from WINE. :)

If time and memory allows me, I will write a small set of terminal control/escape character, so that I might be able to use it as a console connected to a Linux Serial Port.

29 Apr 2000

Sat, 29 Apr 2000 23:38:44 GMT

Today I started to work on a new server daemon for CMC.
It will keep track of the mail accounting.
It will process procmails logs for accounting instead of scanning every mail folder. This way CMC can go really lightweight.
This way I can have more "CMC" apps running. And they all query the server for the info.

The big quesion is: Should I have one daemon per user or a big one handling every users request?

Well, time will tell....