Long time no blogging... teh, teh. Oh well, here's what I have to say today.
There was a Haifa Linux Club lecture a while ago about Ingo Molnar's O(1) scheduler. I got a ride there with Ori Idan. On the way there we ran into a traffic jam and so we were late. The lecture was very nice and interesting, and I think I sort of understood how the O(1) scheduler works. I found out someone was driving back to Tel-Aviv, but his car was full and I did not really know these people, so I decided to go back with Ori (who dropped me only on Hertzelia) as well. Ori and I talked with each other on both ways on some interesting subjects. (too numerous to mention).
Then there was a Tel-Aviv Linux Club meeting. It was about Apache 2. Eddie Aronovich talked at first about "Why Apache?" and then described how to configure it. I expected the lecture to be about the advantages of Apache 2 over Apache 1.x, but this was not the case. In any case, Eddie is a good lecturer and we certainly enjoyed it, and had some discussions in the middle about Apache, IIS, and other stuff. The time of day of this lecture sparked a lively discussion on Linux-IL in which I bashed the long Hi-Tech work-day that is so prelevant.
Helping my sister with UNIX
My sister, Noa, now takes the Technion "Introduction to System Programming" course in which they are required to test their exercises on a UNIX system (a Solaris box to be exact). To help her get along with it, I tried to teach her some UNIX shell, vi and ncftp basics. She did not took my advice to prepare everything on UNIX from the start, and instead did the exercise with MS DevStudio. But naturally she then had to test it on the UNIX system.
So I guided her through that as well. What made matters more complicated was that FTP and SSH access to the designated host is blocked by the Technion firewall and we had to login and transfer files through an intermediate host. I tried my best that instead of telling her the commands, I'll tell her what needs to be done and so she'll have to recall the right commands herself. I wasn't entirely successful.
I don't think she's fully independant now either, but hopefully she'll get better in time.
I passed Statistics with a score of 57%, which is just above the passing grade of 55%. I'm happy this is behind me. Then I saw that the Thermodynamics grade did not enter the system. After some arrangements with the lecturer, it happened, but quite late.
Then, I had to manage my way through some telephone and fax red-tape to become applicable for the degree (which I hope I can get). Now, it is on the waiting queue to be processed, and the secretary told me to contact her a few days from now. Hopefully, I'll get my degree soon.
Pheew... what past few days. First of all, I continued workin on the patch, after there was some silence about its specification. Working on the operation over WebDAV functionality was very time consuming and involved writing quite a lot of code. At one point I found myself in need of recording the HTTP conversation, and having been unhappy with what tcpdump gave, I used Ethereal, which is a great tool and was just what was needed. This helped me resolve the bug, but gdb was also of help in other places.
I also wrote more tests to test the functionality of the added product. While testing, I ran into a nasty gdb bug, which strangely enough was worse in gdb 6.1 than in gdb 6.0. Apparently printing some addresses or variables produced random output while the program run was halted. Hopefully, I'll be able to reproduce it with the final version of the patch.
Writing the bindings for the custom Subversion protocol was relatively easier, because the protocol was simpler, and was not XML-based. After I was done, it turned out the conventions of the callback were inconsistent with the rest of Subversion. So I redefined the specification to be more so, and had to revamp the code somewhat to accomodate for the change. The new conventions actually made the task simpler, but since the code for the previous spec was already written, I just removed parts of it.
Now it's time for the patch to be reviewed. While one developer reviewed some relevant parts of it, and another commented on what appeared in the log. It was not reviewed entirely. I also have some items on the patch' TODO list which I'd like to ask people about.
A recurring theme, was that I was perhaps not as independant as one can hope and kept bugging people on the IRC. I'll try to be more independant next time. Another issue is that people did not like the fact that I would create a branch for working on the patch on the central Subversion repository. Thus, I used working copy diffs, which were "versioned" according to their patches filenames. I have about 29 such patches on my hard disk so far, and also wrote some supporting scripts.
Now I'm going to faint...
I finished reading Julian Jaynes' "The Origin of Consciousness In the Breakdown of the Bicameral Mind". It was very interesting. Now, I started reading "Feeling Good" again, and now see a lot of things I missed or did not read in the first place (can't tell because I don't recall them).
My payment for one of the O'ReillyNet has entered my account. So, I wrote and mailed a check for my payment for my account on stalker.iguide.co.il to nyh. Afterwards, he was glad to help me, and enlarged my quota. I got the Subversion service back online by compiling and installing everything in my home-directory, and restoring the repositories from their dumps.
Then a problem emerged. There was a hard-disk hardware problem on stalker, it went down for a while, and afterwards most of the contents of the disk were restored (including my files), but now we have less space, until the disk is replaced.
Powerout and its effects on KDE
I noticed that after the computer is shut down while X and KDE are running, many of the KDE customizations get lost. My solution was to restore the kdeglobals file from its backup copy. I now keep a copy of this working file at hand for safe-keeping in that directory. And I have no idea why it happens.
While I was working on a presentation with Quad-Pres, I noticed that it always prepares the "Hard-Disk HTML" version of the lecture after every "quadp render". It is supposed to be triggered only by a "-a" flag. The problem was in a local copy that experienced some heavy refactoring lately.
Today, I fixed it in my local copy. There was a small mixup that it occured in the 0.10.x branch and not in the repository trunk, which took me some time to figure out. (the trunk's code looked fine).
That's it for today. Well, this is what happens when you don't write an entry for about three weeks. Next time I'll try to write more often.