november 30, 2002 tracing paths
you know me and my love of graphing. i was doing some work on a paper i'll be presenting at umeet later in december and at cansecwest in vancouver, april 2003. basically i looked at about 100 different keys used to sign software archives and mapped them back to the strong set, identifyed by dtype.org. this is an undirected graph showing you the keys and their paths. i'll release the paper shortly, this is just a sneak peek. a PS and PNG version of this graphic are available.
november 23, 2002 collection
chris turned me on to the idea of snarfing my own RSS feeds and aggregating them into one site for my one stop headline shopping. it's now on the sidebar at the left. this is just a basic version to see how it goes, hopefully i will be able to tune it more appropriately soon (ie mix the link-o-matic ideas with it). i'll have the minimal source of it up soon, but needless to say you can do better if you start with this devworks piece.
november 22, 2002 bye bye
my sourceforge project 'pedantic' has been removed, i wasn't going to maintain it. the fine, kind folks at sourceforge were most helpful in nixing the project. i have removed it from my list of links and projects, unbound is proceeding nicely.
november 19, 2002 spiced ham
above is a small JPEG of a larger graph i generated last night. i had to pass neato a lot of options to get this to look reasonable, and it only worked after i pared down the number of edges to about 1500 or so (down from nearly 3000, and this is my smaller data set). the full thing is available in postscript form as monkey-1500.ps. if you have access to a large scale plotter and want to make me a poster of this, i'd like to know. alternatively, i can probably make you your very own spam based wall art.
the one from my cwru email accounts, cwru-neato.ps, is the full 2200 edges big. i think the smaller number of hops at the extremes made this one easier to graph. and yeah, the blue node (my prime mail server in that area) is under there somewhere ...
november 18, 2002 mirian the librarian
tonight i released the first C library code i have concoted. i have written some other stuff in the past (ganking from other people's stuff or some quick, useless bulk cruft), but this is the first time i have ever shared it.
int cli_loop(int fdin, int fdout, const char *prompt, cli_handler callback)libcli is the start of a small library to assist in the rapid development of command line tools and applications. it has a nice arbitrariness to it i like. works on freebsd and openbsd. version 0.1 is up, and i'm adding to the API as i need to (and think about it, as well). enjoy, and i hope someone finds this somewhat useful.
in other news, i have had neato crunching on some graphs for hours. seems my spam maps have gotten too big ...
november 17, 2002 fault injection
when niels came up with systrace, one of the things dug mentioned was "you could use that for fault injection testing". well, i took a bit and made up some code to randomly modify an existing systrace policy and insert failures. the goal is to run systrace -A app ... exercise it fully, then modify the policy with failgen. here's an example using my silc client policy:
$ failgen.awk usr_local_bin_silc 0.09 > mysilc $ systrace -a -f mysilc silc -c silc.silcnet.org -p 706 $ failgen.awk usr_local_bin_silc 0.09 > mysilc $ systrace -a -f mysilc silc -c silc.silcnet.org -p 706 $ failgen.awk usr_local_bin_silc 0.09 > mysilc $ systrace -a -f mysilc silc -c silc.silcnet.org -p 706all of these worked, i was able to connect and run just fine. well, actually one seemed to have hung for a while, i think it silently failed something. so i generate some more (with higher numbers of failures):
$ failgen.awk usr_local_bin_silc 0.15 > mysilc $ systrace -a -f mysilc silc -c silc.silcnet.org -p 706 Bad magic: ld.so $ failgen.awk usr_local_bin_silc 0.15 > mysilc $ systrace -a -f mysilc silc -c silc.silcnet.org -p 706 No ld.somore errors, but gracefully handled (by the system, not the app). and some more ...
$ systrace -a -f mysilc silc -c silc.silcnet.org -p 706 $ failgen.awk usr_local_bin_silc 0.15 > mysilc $ systrace -a -f mysilc silc -c silc.silcnet.org -p 706 /usr/libexec/ld.so: silc: libperl.so.6.1: No such file or directory Memory fault (core dumped) $ silc --version Irssi-SILC 0.9.4 (Irssi base: 0.8.5+ - SILC base: SILC Toolkit 0.9.2) (20020623 20020623)and we have success, we broke our app (in at least one spot). the systrace policy which generated these errors, is in the examples directory. failgen itself is available, too. this is an early prototype. much thanks to dugsong for inspiration.
current problems in faultgen are at least the following:
later on ...
released version 0.2 of failgen, which fixes a problem where spurious core dumps would get caused. if you deny the system to exit, it has to quit somehow :)
november 11, 2002 suckage
most blogs suck. this site sucks, i hope you come here only for the links to the quick widgets and code i write. instead, you should read lynn's site. when people want to do websites in journal format, they usually think they can be as literate as she is. instead, no, we usually fail. humbling, humiliating realization.
i crave sleep. haven't slept well in a while.
i'll get a snapshot of unports up hopefully tonight. been hacking on unbound some this weekend. cleaned out a few developers, added one, and did some checkins. code's kind of broken for crypto, which sucks. need to dig into it and fix that.
new snapshot up, unports-11112002.
november 10, 2002 digital photos
did a port of tls, the tls extensions for tcl, to openbsd yesterday. it will be in the unports snapshots soon.
beth is getting me thinking about a nikon digital slr. they finally got cheap.
fixed up picpages. it now supports embedded copyrights (via outguess) and a real setup option :) version 0.9.1 is now up and available.
november 8, 2002 too busy!
silly me. cyclone-0.3 doesn't operate on anything but x86. so, no, you can't use it on PPC or ultraSparc. *shrug*
so many things i want to do ... here's a list of things i want to port to openbsd real quickly (for varying degrees of quickness). been dorking around with mono, the gnome c# stuff. it doesn't yet work on openbsd ... but hopefully soon. jmallett has positive experiences on her openbsd-macpps laptop, so i will keep poking at it. plus i want to start dorking around with mod_dav and see if beth can't play with a .mac server at home that doesn't rip her off. (apple has charged her something like $250 now after saying they were closing her account. whoops ... apple, you suck.)
need to contact someone and see what's up with the soon to be released book.
november 6, 2002 pears are for eating
pasta with pears and gorgonzola
2 firm and ripe pears
peel the pears and core them. slice them into 1/4 inch thick slices. saute them on meadium heat in the butter until they start to brown a bit. as you remove them from the heat, toss them with the crumbled goronzola. toss this immediately with the pasta. serve warm. if you would like, you can saute the pears in the butter with a dash of pear liquor.
pasta with vegetables and pesto sauce
12 oz fetuccinni or similar pasta, cooked al dente
saute the zuchinni and squash in the olive oil over medium heat, toss the pasta with the pesto and vegetables. serve ... :) easy meal.
beel drooling over fogbugz lately. been looking at how to develop something similar under a bsd license ...
november 5, 2002 cyclone port is done
finally finished up a port of cyclone 0.3 to openbsd. it will definitely be in the next unports snapshot. if you're interested in trying it out on other platforms, please get a hold of me.
other than that, just busy doing the same old stuff: writing, working, hanging out with beth. *hrm* we forgot to drink some hot cocoa the other night, we'll have to fix that.
november 3, 2002 new stuff is up
new snapshot of the unofficial ports tree is up: unports-11032002.tar.gz. includes some updates as well as some new stuff: cflowd, arts++, and the like.
new pics up, too, see the sidebar to your left.
november 1, 2002 lots of graphs
been hard at working making graphs. the first is spam. you guessed it, i graphed some more about my spam. i decided to look at how much spam i get per day. turns out its not nearly as regular as it feels some days. this is the number of pieces of spam received per day (plotted using the system's received on date stamp).
the second is a set of system call graphs (syscalligraphy, a term coined by jeff). they're all in postscript (deal): ]if you want to make your own you can see how. they're useful for just analyzing a bit about your system's complexity. thanks to niels for suggestions on using ktrace a bit more effectively. i need to group these syscalls together, now, into subgraphs (ie for sh, sed) in big things like ./configure.