bleh! resnet still isnt up, and i'm bored at uni.
build secret tunnel to stirling, break into mieks flat and steal bubblegum crisis tapes 4-8. i gave up at number 4 last night about 6 am. (shame)
just saw a very nice web page on running a keysigning party got me thinking, maybe I could arrange that at my next LUG. then again, maybe I should go the full geek and get my GPG Fingerprint on a tshirt :)apache
If i ever find the need, I will patch apache so it produces XHTML valid code, instead of the broken HTML3.2 it generates now :/
this is something I mentioned to peen and rt, but didnt post. for a webblogger, when adding a post, add it to the front page *and* the archive at the same time. so when a post is too old (on the front page) you dump it. simple :)
more specifically, the post script copys the news into two files, news/<unix time> and appends it to archive/<month>-<year>.html. and then removes the oldest entry in news. the front page generator the latest 10, or whats on in the last week (depends) and removes the old news items. el nifty.
In case you don't know me, im /evil/ (and i have a whole channel of idlers to prove it), anyway just had an idea for email worms (don't believe write them [then again, when have I ever had the effort to code] , i find a hatchet is a more efficient way of getting revenge [bwahah]) most email virii make up a random subject and body of the text file, or piece together a sentence from words in an array. I thought it would be more efficient to do grab a rss file from your favorite news site (possibly a random rss) and use that to generate the subject/body (fools detection algorithms), and then attaches more-info.(evil extension).
not that i'm a vx'er or anything but sometimes virii are coded better than the platform they infect (unless they're written in vbs *eugh*). another evil idea i had was parsing the inbox of infected host and then randomly replying to mail with some odd reply with the usual attachement. (my-thoughts.doc) etc.
enough evil for now. idling is calling me.
hmm after much bickering^H^H^H^H discussion with xs the resident security bigot of #argv it turns out there isnt a patch to assign root capabilities to specific gids (and then kludge setgid to set egid like it does for root as they are technically root privs). shame. *sniff*. and I'll just put it on my big TODO list.
Ended up talking to mike about filesystems, basically he's being using resier to avoid fsck on a webcache. (don't blame him though, I should be using XFS soon on /home). fsck is a problem on a 36gb system apprently.. so I got thinking about a temporary filesystem where data recovery is non existant. The idea instead of running fsck on the fs when rebboting after a crash, it would be quicker to mkfs then mounting. (this is why it's called tempfs) quicker yet would be a dd copy of a nice clean superblock). If I was serious about this i'd patch a copy of ext2 for speed (no updating timestamps etc), and have the stickybit permanently in effect. (so it could be used for tmp and a cache...). It would probably write a clean root dir when it mounted perhaps. and of course if you wanted to recover the data after a crash you could always fsck it and mount it as an ext2 drive. (Isn't technology wonderful)weird
got certified as apprentice and I didnt even bribe the people :)
Hmm. I Was pondering more on steg fs'. Probably having a bunch of 1k files in a directory containign random data. you parse each one , decrypt it and check the hash/crc at the bottom to see if it's your file. any files that cant be decrypted would probably be declared as freespace and overwritten. its a bit of a duct_tape hack. perhaps they wouldnt be shown. then again. it has to be a pretend ext2 system allegedly.
urk! forgot to mention that there isnt a linux distrobution for lazy people. guess i'm just too much of a slacker
maybe debian could be adapted, i'd rather have some sort of meta packages rather than all in one .deb (or whatever files).
lazy people need a distrobution, to be honest. I can't be arsed making one
Was originally going to host this on nitric (see above) but advogato is quicker easier and I dont have to worry about fixing The bugs.the irc client i migth just finish (or start..)
so what follows is a few developmental ideas that have been running aorund in my head for a while...
The irc client i'm 'developing' is still a blotchy picture in my mind, i'm still not sure what language to start it in as I know perl, but would like to do a little more than 'hello world' in c...paranoid file systems..
the irc client. i't would probably most resemble a instant messager client pager containign all the current iwndows and status, and then mirc-alike windows for channels and queries and crap like that.
I was rethinking about stenographic file systems e.g (stegfs) and i've been playing around with an implementation based on ron rivest's chaffing system of hiding data in plaintext, the idea being having all inodes on a system being valid and you'd have to check a md5 sum (included) or some such. this would also be true for indirect inodes.. (so even if a inode is correct , it might only be 10 blocks big and the indirect inodes might be false)... The hardest bit is the maintaing of the filesystem. then again there could always be a copy hiding on the ext2 partition...more unmade code
although security bigots would crucify me for this, i wouldnt mind a /proc entry showing the last 10 or 20 deleted files... (especially as i've regretting rm'ing a file or two..)
another idea i've yet to implement (you /will/ see this a lot on these pages) was a news whore script. basically i'm pissed off with the news sites with the bloated designs (if it doesnt work in lynx, its bloated). like *cough* slashdot */cough*. So i was then thinking abour parsing the front page of a couple of sites. sf pointed me at rss (but it doesnt have everything). indymedia would probably get treated too. (even if it is fast, well written, and informative)
anyhow, less wishful thinking and more code should sort me out :)
(and one day i'll do that linux from scratch I was always going to do)
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!