You know what I love about weblogs? People can say whatever random things they like and pretend that they are correct. And maybe even that people listen. You can even just make stuff up.
In some ways, it's a lot like writing for Wired.
More new stuff on the system tuning page. apache mod_proxy useage, some tcp tuning stuff, listenBacklog. But more importantly, I've started trying to merge the info there to linuxperf.nl.linux.org.
Slowly getting back into the mood for coding again. Coming off some deadlines, and then last week, kind of threw me for a loop.
In the meantime, been updating my system tuning page again. Added some info about NFS tuning, disk io elevators, pty/ttys, and misc network settings.
Ever get in the mood for a pointless long diary update? I do.
work
Life
It is way too hot outside
All excellent
Looks like some coworkers ( taw, misa), have joined advogato. Of course cturner, DV, and jbj were already here.
Work proceeds as normal. Lots of deadlines. Lots of python coding. More and more SQL. More and more pygtk/glade.
Progress continues on up2date. Spent most of last week fixing a few dozen bugs in it, mostly minor thankfully.
Btw, if your writing python code and not using pychecker. Run now and install it and run it on your code. Very handy. I've taken to adding pychecker targets to our makefiles that build python code.
Looks like the eternal package format wars continue. Software packageing is hard, lets go shopping.
Went and saw Lake Trout a few days ago. Highly recomended. Odd crowd though. Apparently they were sort of a jazzy hippy jam band at one point, but then got sort of electronic, and then lots heavier. Seeing a crowd of about half "hippies" doing that spinny dance thing to the encore of "War Pigs" was interesting, to say the least. Great show, made even better by the fact that my friends' band "Japan Air" opened up and played well.
Also saw the movie "The Score". Not bad. A little predictable, but okay. The "hacker" in the movie included a stuffed penguin on his desk, and at some point was shown typing away at a bash prompt after logging into a "linux 2.0.30" box. 2.0.30? Did they find a copy of 4.2 in a timecapsule or something?
A coworker brought in a five pound block of silly putty
If you have ever been curious what five pounds of silly putty looks like: pic1, pic2
I'm not if it was the cold I currently have, or the silly putty that caused today to be mostly unproductive.
Slowly getting used to playing the Chapman Stick. It's both easier and harder to play than I expected. Of course, being pretty much tone deaf doesnt help any.
Speaking of being tone deaf, hacked around a bit on GNU Solfege. Mainly just to add support for Chapman Stick in addition to the already existing piano/guitar/base/accordian support. Need to submit the patch here shortly.
Found a picture of Patrick Stewart playing a modified Chapman Stick in Dune. Thats too many geek themes in one picture for my poor little mind to handle.
RHN is keeping me busy. Nice to see people actually using it. I've actually used up2date to update a 6.2 box to a 7.1 box just for kicks, and a couple 7.0->7.1 upgrades. Kind of nifty. Not really "supported" at the moment, but interesting none the less.
Made the mistake of blindly buying a new "Sound Blaster 16" sound card without reading the box with a cynical view. I wanted a card with some degree of hardware midi support, even if it did suck. Turns out the new "Sound Blaster 16" is just yet another relabled ess1371 card. Great. I'll add it to the pile. Borrowed a SB awe32 for the time being.
Do not ever wander into music stores not expecting to find anything interesting. I made the mistake of doing that last weekend, and seeing something interesting.
I really never had any urge to buy one before. But then, I had never seen one in a store before.
So I thought about it for a week.
I bought a Chapman Stick
It's cool. Everyone should have one. Follow the link if you do not know what a Chapman Stick is.
Or take a look at the pictures of mine.
Of course, I can't play the thing quite yet. Sill kind of in the stages of figuring out how I want to tune it. It's a very old (July 76') 10 string. This should be interesting.
The fact is..
no matter how closely I study it,
no matter how I take it apart,
no matter how I break it down,
It remains consistant.
I wish you were here to see it.I like it.
--Indiscipline, King Crimson
In the software world, just more of the same old rpm/python/up2date hacking.
re: changelogs and cvs2cl/rcs2log
Autogenerated Changelogs are probabaly more than adequate for projects with one or two developers, but for larger projects (say, the Gimp), the extra context that a hand written changelog provides can be very useful.
Namely because hand written changelogs tend to document what the author intended to do, where cvs logs document what you actually changed. Especially useful in the "did adrian forget to commit all the new files again?" scenario.
Rather or not that is worth the effort is another question.
Been a while. Oh well.
Anyway, looks like a lot of people have expressed interest in how we do xml-rpc over (under?) ssl, and a lot of speculation. Of course, the answers are in the source code ;->
Basically, there is a https class hacked into the python package based on the M2Crypto openssl bindings (in the openssl-python package). Then some mods to python-xmlprc to allow the use of https and to do CA checking. This is what the client apps use to do ssl
Server side is apache and mod_ssl/openssl.
Seems to work well.
As far as the question of "who needs more than 4gig of ram", the answer is apparently "lots of people".
It was a pretty common request while I was working in tech support. As often as not, it was people looking for >4gig support per process, and at least with 2.2, x86, and linux, that wasnt really an option. But it wasnt uncommon for someone to want/need to be able to malloc a half a dozen gigs of ram. And not just malloc it, but use it as well.
But very often, people just needed more memory available because of huge numbers of proccess running. Web servers running on big hardware with long lasting cgi/asp/*let proccesses were a common theme. Ie, a few thousand perl processes taking a couple of megs each on a single machine. Yes, people really do that.
Or perhaps, you just need to serve up several thousand hits a second with a single machine. TUX and 32gigs of ram to the rescue... Okay, so I dont know anyone actually doing that in production. But one thing I learned in tech support, is if there is a limit on something in linux, someone will run into it.
New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.
Keep up with the latest Advogato features by reading the Advogato status blog.
If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!
