hpa and I finally got around to upgrading the master kernel.org hardware to a dual 1666MHz athlon from the previous single 1400MHz athlon (formerly a dual with one thermal failure). You can tell a difference, but BK is still a big pig and the system could use another 512MB of ram at least.

Well, I spent all day with the master kernel.org server failures.

It turns out that one of the Athlon fans died, so it's now down to one cpu, and two of the hard drives died, and one of them was wedging the 3ware linux driver. Replaced both of those drives (one of them was /dev/sda so there were boot issues to fix).

At about 9PM I had everything sorted out and I wondered how hpa knew the exact right time to skip town...

Well, I'm down to 5 warnings remaining on the stradis.c fixes for linux 2.5.x. It should probably work in its current state, but I want to limit what I can point the finger at if it doesn't work on the first try to poorly placed locks rather than bad type casting...

Started a 2.5.55 compile today in preparation for moving kernel.org to 2.5.x and making linus eat his own dogfood.

Of course, the compile failed, so I grabbed 2.5.58 which also failed. Apparently there are fbdev includes missing in the tar file (which I did find on Linus' machine at work, but there were further issues).

After disabling fbdev, i discovered errors in the bttv driver (which i fixed) and the stradis driver (which are so extensive that I'm still in the process of fixing). I also discovered that some people don't know how to write proper gnu syntax asm and changed binutils to accept their bogus code. Fixed.

Ugh. I really wish people would report kernel code that doesn't compile to the MAINTAINER of the code rather than expecting that everyone has the time to sift through LKML looking for the random bug report.

Well, I've given in to peer pressure and created an account and even created a freshmeat entry for one of my oldest projects, playmidi.

I'm still pretty upset that someone decided that /dev/sequencer was useless in newer sound drivers, but I was recently happy to learn that ALSA drivers have all the support needed for Playmidi and should be a standard part of Linux 2.6 kernels...

Also, made some really tasty double chocolate cookies from a recipe in Cook's Illustrated (Page 25 of Sept/Oct 1999) in the wee hours of this morning. It says that it makes 3.5 dozen, but I came out with about 19 cookies, so you can imagine how big these cookies are. Also, I melted a big brick of belgian chocolate and not chips, so there's a lot more cocoa butter, and the taste is quite decadent.