Older blog entries for claudio (starting at number 96)

These PalmOS devices are incredibly wonderful toys. They even can be used as organizers! After a few years considering them useless for my needs, I ordered a IIIc unit and I'm really impressed with its simple, clean, ingenious design. Now I must find a decent way to keep track of my bugzilla tickets using it (perhaps importing them to the To-do list?), with comments as attached notes?

Yesterday was a bad hair day. Went to bed 02:00AM, woke up this morning at 6:30AM with my hair still bad, hopefully it will calm down today.

I'll move xmp to sourceforge this week. And I still can't add igor_nest to the sarien project.

In the past few weeks I'm being much more productive working at home. It seems that the noise levels and the number of spurious interrupts increased noticeably in the office, or my perception of the surroundings has been enhanced somehow. In either case, all these small annoyances and disturbances are slowing down my CPU cycles. Or it's just ADD, and in that case I should take Ritalin and listen to Phil Collins :)

Or perhaps I should get a sound card and a pair of headphones -- at home I've been working listening to things like Primus, The Residents, Kronos Quartet, Astor Piazzolla (!) and Malmsteen's Concerto for Electric Guitar. And some Greek music.

acme should post his article on software packaging. /me takes note: bug him to post the article.

Interesting findings can arise from silly mistakes. As jameson pointed out, in my previous diary entry I made a typo in the Sarien URL in Sourceforge -- and it pointed to an existing project I wasn't aware of!

Using the FreeSCI graphics subsystem in sarien could be a good idea, but before doing anything I must clean up that code. I suspect there are many bugs lurking from the dark corners of that code. BTW, does anyone know if there is any problem using glob() in BeOS? It seems that Sarien built fine except for a missing glob.h.

Regarding FreeSCI, I must check how is the sound subsystem going. I've worked with /dev/sequencer and an Adlib card emulator in xmp.

(Update: fixed the sourceforge project url. Interesting that by mistake I pointed to another project I didn't know, an AGI interpreter in Java.)

Brought Sarien back from the dead, and put the source tree under revision control in Sourceforge. There's a lot of dust & rust to be cleaned up in those lines of code in order to make it maintenable -- several layers of dirty hacks must be straightened out, and old kludges to be correctly addressed. It's also interesting to see how your coding skills have evolved in a couple of years, and how your code standards have changed.

Inspired by jameson's FreeSCI documentation, I've converted the currently existing AGI docs to DocBook and put it in the Sarien package. Lance Ewing's AGI utilities are in the bundle as well. Currently the SGML file is half-converted and under a revision process. A few sections have been removed and will be put back after the rest gets organized.

The QNX Photon driver for Sarien is already in CVS, and must be properly integrated to the config system to be built out of the box, as well as the win32 port via cygwin cross-compiler. I'm also tempted to port Sarien to PalmOS :) The Palm 160x160 screen is almost the correct size for the AGI 160x168 image resolution, I just wonder if the IIIc's CPU will be fast enough to run it (the Visor Edge is faster, but we'll need a colour display). Hmm, another reason to buy a PDA ;)

Finished my M.Sc. thesis, presented it. MagicPoint did a great job allowing me to embed the dynamics simulation window directly into the slide. The algorithms performed quite well, and I'll probably use them in a hack for jwz's xscreensaver. When I get time, that is.

My advisor was quite pleased to know that I had everything ready for presentation a several weeks before the deadline, and that I praticed it until it was perfect and all. Well. Actually I prepared the slides the night before the presentation :) That's the way it works, if you know what you're talking about.

Lots of things in my todo list. First of all, I'm resuming my work with USB in the linux kernel. I've been away from the usb-devel list for some time, and now I'm trying to get in sync with the current state of the project. Previously I did some work on Mark McClelland's OV511 camera driver, and hacked the parallel quickcam driver a little bit.

In the apt front, it seems that the byte vs. kbyte bug in the progress indicator is there again. It doesn't affect apt-get too much (except for the wrong values reported), but aptitude gets very unhappy with this problem. I believe this will be very easy to fix.

A very nice parallel service manager was written by epx, and I'm tweaking it to ensure full compatibility with traditional SysV initialization and at the same time get advantage from threading. People who think their systems take too much to initialize probably will like it.

Some technical papers and articles are patiently sitting on my filesystem, waiting for completion.

I also think it's time for me to buy a handheld such as a Palm or Visor. I used to make notes in pieces of paper, post-its, old tickets and the like. The Visor Edge looks like a nice unit, with its 33Mhz DragonBall and thin design. Too bad it's not for sale in Brazil :( Also in my wishlist is a digital camera -- perhaps a Sony Cybershot? Well, I'll try to find a Visor or Palm first. My primary target is a fast BW handheld, but a color unit would be a nice platform to port the Sierra AGI interpreter to! :)

Also must talk to my MSc advisor to buy his old P133 mobo and processor, to replace my P100-based DSL firewall. Not a big deal in the CPU itself, but I certainly will see the difference between my current 20Mb of RAM to 64Mb.

And, just to finish this entry with my favourite phrase, Brasil Telecom employees are idiots :)

Ah! Finally finished that robot dynamics thesis. I'm back to life! No more Recursive Newton-Euler and Composite Rigid-Body Method! Bwahahaha!

Erm.. wait. I still have to present it and stuff. Blah.

You are on the way to destruction!!

Notes regarding the past few days:

• Got a new machine at Conectiva and installed a fresh 6.0 (the old machine collapsed , disk AND mobo). GRUB is a bit strange at first but I'll get used to it (or so I hope). I installed only the essential packages and got the rest using APT. The new machine has an OHCI controller, so I'll finally be able to play with the OHCI drivers.
• Writing the dissertation. Found an error in the previous data sets, so I ran all the experiments once again.
• Tried FreeSCI and XRally on the PowerPC machine.
• Updating Conectiva's aptitude RPM to 0.0.7.11.
• DSL kicks ass. It's not exactly fast, affordable or reliable, but it's way better than dial-up! :)
• The dissertation deadline approaches. Don't panic. Don't panic.

Notes regarding the past few days:

• The DSL link is fine now. Of course the telco will now make external modems available for half the price I paid for mine, but who cares.
• Back to USB devel. Now reading docs and getting in sync. Maybe we can build a cheap protocol analyser using the host-to-host cable and a windows box.
• Thesis. Rewrote a chapter, reorganizing stuff. Chapters 1, 2, and 4 are pretty much done, must finish 3 and plot the graphs for 5. Chapter 6 is almost complete too.
• Playing with QoS and sch_cbq. Let's see the difference it will make.
• Fixed intro/hiscores music in XRally.
• Met some old kindergarten classmates :) Ouch, they're really OLD!

After focusing all my efforts on Conectiva, and to put my DSL connection to work, it's time to finish my thesis once and for all. It seems that DSL is finally working as expected (well, sort of, I'm still experiencing some frequent LCP timeouts and stuff. But it works! Oh boy.).

So, the thesis. Finish it. Collect data. Write. Once and for all. Hmm... new South Park season... No! Thesis. Thesis. Finish it!

Interesting news from the DSL front (and it seems, from wainstead's diary, that DSL is sort of a nightmare everywhere).

So my DSL link stopped working, and I called tech support again. They came to measure my line with a lot of nifty gadgets and declared it perfect. To my surprise, my modem actually sync'd after the test, and the tech guy smiled in the "I told you, stupid user" way. BUT -- it stopped working after the first modem reboot. Hmm. Only then they actually believed thet "may need to check" their perfect equipment. Telcos are stupid everywhere *sigh*.

The diagnostic is that Lucent Stinger DSL concentrators doesn't interoperate well with 3Com Homeconnect Dual Link modems -- no comments about the Lucent and 3Com Corporation design specification for ADSL interoperability, which seems to be vapour, like the 3Com concerns on interoperability (either that, or those telco guys are too stupid to know how to configure their own equipment):

In the effort to make interoperability a reality 3Com has announced agreements with several central office equipment vendors. These agreements call for 3Com to perform a variety of tests to ensure interoperability between the central office and 3Com's customer premise equipment. Current ADSL interoperability partners include Advanced Fibre Communications (AFC), Alcatel, Diamond Lane/Nokia, Lucent, Nortel, Newbridge and Redback Networks.

And they say more (heh, this part is kinda funny!):

3Com strives to achieve interoperability with all leading central office equipment vendors. The ultimate goal is to allow anyone who is qualified for ADSL service to use 3Com ADSL Modems at their home or office. It won't matter what equipment is at the central office'a 3Com ADSL modem will "talk" with it.

For some reason I'm believing more and more in the fact that the industry isn't able to follow standards. They're too dumb for that.