Ten Thousand Statistically Grammar-Average Fake Band Names
Some nice ones:
And noticed more than a few existing bands: finger, breadwinner, giant, iodine,
acetone, etc
Sinus headache last night, think I took one too many nyquil tablets.
Conked out of work a little bit early, went home, couldn’t sleep.
Hooked up the mixer and a couple of pedals and made
some noise Merzbow style. The digitech muliplay (modulating
2 second delay) into digitech rockbox (chorus/distortion)
into a yamaha 10 band eq. The input of the chain was
the effect send on the mixer, and the output fed the input
of the mixer. ie, a feedback loop. And just for kicks, the
Vortex was in the second effects loop. With a little
experimentation, lots of fun noises can be made. (no
no actual instruments in that setup… fun!)
More up2date work. Cleaning up ui issues now.
So far, haven’t accomplished anything this weekend. Which sucks, because
I really needed to. Suppose I should at least buy some groceries and what
not.
Picked up a copy of Guitar Player magazine yesterday with the
cover story of “The 50 Coolest Stompboxes Ever”. I seem to
only own one item on the list. Oddly enough, the Snarling
Dogs Wah/Ring Mod made it.
Hmm, closed a couple of dozen bug reports, and updated about
60 bug reports in total. And got “up2date –undo” working
to some degree. Not a bad day productivity-wise.
Browsing the web isnt showing me any interesting shows
coming up this weekend. 2/3 of Morphine at Cat’s Cradle
could be interesting. And my friends The Finks and
Wafer Thin are playing ps211
saturday. But not much else. Probably all the better since
I have a lot of work to get done.
Plodding though bugs. After lots of profiling, I’ve mostly decided that up2date
isn’t actually all that slow. Turns out it was averaging just under a
second for each package, and just about two and a half second
for each package that was repackaged. Of course, that looks
really slow when your updating 1500 packages. The numbers look
even better without “dev” or “glibc-common” or “kernel-source”
throwing the curve (15-25% of the files on an average system
are in those three packages…)
Left work somewhat early (9:00pm ish…) to head over to
Durham to catch a show at Ringside with badger.
Just your typical tuba/cello/bass/guitar/bass clarinet/moog
improvisation group ;->
Got home and actually played a bit since I was in a
“make (hopefully)interesting improvised noises” mood. Of course,
thats pretty much all I ever do.
I need a good “Reply-To flamewars considered harmful” url.
While attempting to profile this stuff, I managed to create
a profile data file that includes a time delta of “-1” which
of course, breaks the stat generation modules.
Awesome, I cant even profile correctly!
testers are complaining that dist upgrades with up2date are slow.
And they are.
Part of this is new rpm stuff. Enough of it that its
hard to see what is and isnt.
But other than that, I need to make it fast.
Considering it:
– downloads about a gig of data
– downloads that data over ssl
– stats basically every file on the system
– performs somewhere on the order of half a million md5sums
– runs gnupg on about 1000 packages (about a gigs worth)
– gunzip/bunzip2’s about a gig of data
– performs lots and lots of DSA verification of headers
from the db (most of this is turned off, but still)
– cp just about every file on the system about 3 times
(once to a backup, once to put it on the filesystem, once to
build the repackage rpms)
– rebuilds about 1000 packages (this a completely
new unoptimized code path by the way…)
– gzip/bzips about 1000 packages (about a gig)
– solves a dep graph that can consist of about
1500 nodes and somewhere about 30,000 edges
That should just fly by!
The good/bad news is that all my profiling so far points
to rpm just being slow for the bits that don’t fall into
the above. Grumble.