<kristianpaul>
i just need a tilling window manager with same screen keybiding and that allow me to take a session remotelly just like screen does, is too much ask? :-)
<roh>
uh.. spares of spacers... mmmh. no clue to be fair. i had every single one in my hand and checked the outer threading..
<roh>
not sure anymore if i checked the inner one
<wolfspraul>
roh: funny, eh?
<wolfspraul>
it's amazing that over time (with enough volume) - EVERY POSSIBLE thing that can go wrong will go wrong :-)
<roh>
true
<roh>
had some weird stuff in screw-bags already
<roh>
some didnt even look like a precursor to a screw
<wpwrak>
pessulus neanderthalensis ?
<wpwrak>
or, rather, clavus neanderthalensis. (since it's not a screw yet)
<wpwrak>
the current corruption run is nice: 1452 cycles and counting. it's nice because it makes the empirical distribution fit an exponential distribution better. i'm beginning to see order in the chaos :)
<wolfspraul>
wpwrak: and what's your gut feeling now?
<wpwrak>
it's still too early to tell. this is the statistical baseline. with that, i can now see if changes have a quantitative effect.
<wpwrak>
one problem with getting the statistical baseline is that the probability of a cycle causing a corruption is about 1/500. and each cycle takes about 80 seconds (in the conservative configuration). so it takes a lot of time to accumulate enough data. now i have 12 data points, which already produces a reasonably clear curve.
<wpwrak>
so once i'm done with the new scheduler, i'll switch to a new test pattern and see what happens
<wolfspraul>
hmm
<wolfspraul>
ok :-)
<wolfspraul>
not much I can say I guess
<wolfspraul>
Adam keeps fixing & testing
<wpwrak>
in any case, whatever i find, this won't have much effect on rc3. when do you think work on rc4 will start ?
<lekernel>
most of it is DIY and made of junk parts
<lekernel>
I particularly liked the spot welding machine based on a rewound microwave oven transformer, and the tube spinner (to melt the glass uniformly) based on a vinyl turntable