2008-09-07 00:24 <shapor> flips: ping
2008-09-07 00:24 <flips> pong
2008-09-07 00:24 <flips> hmm, my autoponger seems to be working
2008-09-07 00:25 <shapor> heh
2008-09-07 00:25 <shapor> so, tux3fs isn't working for me now
2008-09-07 00:26 <flips> whoops
2008-09-07 00:26 <shapor> i upgraded my fuse
2008-09-07 00:26 <flips> let's see if it works for me
2008-09-07 00:26 <shapor> but when i run tux3fs, it just hangs
2008-09-07 00:26 <flips> gdb is your friend
2008-09-07 00:26 <shapor> # ./tux3fs /tmp/testdev /tmp/test -f
2008-09-07 00:26 <shapor> devmap_blockio: read [2]
2008-09-07 00:26 <shapor> devmap_blockio: read [3]
2008-09-07 00:26 <shapor> lookup inode 0x0, 0 + 0
2008-09-07 00:26 <shapor> mode 0000000 uid 0 gid 0 root 4:1 ctime 0 size 20
2008-09-07 00:26 <shapor> lookup inode 0xd, 0 + d
2008-09-07 00:26 <shapor> mode 0040755 uid 0 gid 0 root 8:1
2008-09-07 00:26 <flips> is this make testfs?
2008-09-07 00:27 <shapor> well yeah, but run manually
2008-09-07 00:27 <shapor> due to no sudo being installed
2008-09-07 00:27 <flips> ls -ld /tmp/test
2008-09-07 00:27 <flips> drwxr-xr-x 0 root root 0 Dec 31  1969 /tmp/test
2008-09-07 00:27 <shapor> although, actually, it does work
2008-09-07 00:27 <shapor> i just have to be root to see it
2008-09-07 00:28 <flips> so what is the hang? <- joke
2008-09-07 00:28 <flips> I see
2008-09-07 00:28 <shapor> as a regular user i get
2008-09-07 00:28 <shapor> ?---------  ? ?      ?            ?            ? /tmp/test
2008-09-07 00:28 <shapor> heh
2008-09-07 00:28 <flips> right
2008-09-07 00:28 <flips> until somebody fixes it
2008-09-07 00:28 <shapor> ok i'll fix it
2008-09-07 00:28 <flips> I'm back to adding new grooviness to tux3 itself
2008-09-07 00:28 <flips> thanks
2008-09-07 00:29 <flips> post on the version table coming in a few minutes
2008-09-07 00:29 <shapor> just had a bunch of caffeine, i'll be up a while
2008-09-07 00:29 <flips> heh
2008-09-07 00:29 <shapor> fun
2008-09-07 00:29 <flips> I just played them demo of disney's new sick trick quad game, I'm kinda hyped
2008-09-07 00:29 <flips> it's good
2008-09-07 00:29 <flips> recommended
2008-09-07 00:29 <shapor> video game?
2008-09-07 00:29 <flips> ridiculously over the top quad bike racing/tricks game
2008-09-07 00:29 <flips> yes
2008-09-07 00:30 <shapor> eh
2008-09-07 00:30 <flips> you'd like it
2008-09-07 00:30 <flips> it's sick
2008-09-07 00:30 <flips> not much gore though
2008-09-07 00:30 <shapor> i can't get an adrenaline rush from video games anymore
2008-09-07 00:30 <shapor> kinda pointless
2008-09-07 00:30 <flips> probably would from this one
2008-09-07 00:31 <shapor> if i want to entertain myself i can go and do stupid shit in real life
2008-09-07 00:31 <shapor> .. the advantage of having a 180hp motorcycle ;)
2008-09-07 00:31 <flips> you can watch this, then go out and kill yourself completely
2008-09-07 00:31 <flips> it's pretty sylish
2008-09-07 00:32 <flips> funny thing is, people are actually doing tricks really close to what's inthe game
2008-09-07 00:32 <flips> the game was meant to be ridiculous
2008-09-07 00:32 <flips> well
2008-09-07 00:32 <flips> real quads don't have helicopters flying beside them at the top of a jump
2008-09-07 00:35 <shapor> they could
2008-09-07 05:10 -!- stargazr5(~gauravstt@59.95.24.129) has joined #tux3
2008-09-07 05:14 -!- cdk(~chinmay@121.246.36.66) has joined #tux3
2008-09-07 05:19 -!- cdk(~chinmay@121.246.36.66) has left #tux3
2008-09-07 06:56 -!- dipanjan(~chatzilla@122.167.27.144) has joined #tux3
2008-09-07 09:42 -!- dipanjan(~chatzilla@122.167.27.144) has joined #tux3
2008-09-07 11:19 -!- pgquiles(~pgquiles@253.Red-83-44-239.dynamicIP.rima-tde.net) has joined #tux3
2008-09-07 15:56 -!- MaZe(~MaZe@c-24-6-86-168.hsd1.ca.comcast.net) has joined #tux3
2008-09-07 16:11 -!- tim_dimm(~timothyhu@cpe-76-90-98-247.socal.res.rr.com) has joined #tux3
2008-09-07 16:13 <flips> another big design note up
2008-09-07 16:14 <flips> in case anybody thinks these take longer to read that to write, I can assure you that is not the case ;-)
2008-09-07 16:16 <flips> http://slashdot.org/comments.pl?sid=954803&cid=24892667 <- warm n fuzzy
2008-09-07 16:20 <shapor> flips: i still don't see your design note
2008-09-07 16:21 <shapor> ah just delivered..
2008-09-07 16:21 <shapor> slow
2008-09-07 16:22 <shapor> flips: i emailed a patch, and actually just commited a fix for a non-bug (yet) in it in my repo
2008-09-07 16:24 <shapor> actually this is dumb:
2008-09-07 16:24 <shapor> return inode ? 0 : -1;
2008-09-07 16:25 <shapor> return -same as:
2008-09-07 16:25 <shapor> return -!inode;
2008-09-07 16:27 <konrad> you mean -!!inode?
2008-09-07 16:27 <konrad> oh, nevermind
2008-09-07 16:30 <flips> back
2008-09-07 16:30 <shapor> flips:  how many bits for atom numbers?
2008-09-07 16:30 <flips> variable
2008-09-07 16:30 <flips> start with 16 bits for small ones
2008-09-07 16:31 <flips> and have a bigger variant
2008-09-07 16:31 <flips> with say 48 bits
2008-09-07 16:31 <flips> or 32
2008-09-07 16:31 <flips> don't have to get silly
2008-09-07 16:31 <shapor> this may be painfully obvious
2008-09-07 16:31 <shapor> but, why limit that approach to just xattrs?
2008-09-07 16:32 <flips> because nearly every file has a data attribute
2008-09-07 16:32 <shapor> i was talking about this for sharing user, group, mode
2008-09-07 16:32 <flips> and mutliple versions of it
2008-09-07 16:32 <flips> ah
2008-09-07 16:32 <flips> of course
2008-09-07 16:32 <shapor> bundle them all together
2008-09-07 16:32 <flips> what I meant in my post
2008-09-07 16:32 <flips> how bundle?
2008-09-07 16:33 <shapor> i mean, put the user, group, and mode in the atom table too
2008-09-07 16:33 <shapor> along with xattrs
2008-09-07 16:33 <flips> sure
2008-09-07 16:33 <flips> good idea
2008-09-07 16:33 <flips> separately or combined?
2008-09-07 16:33 <flips> the latter gets into content addressing
2008-09-07 16:33 <shapor> probably combined
2008-09-07 16:33 <shapor> more efficient that way
2008-09-07 16:34 <flips> more corner cases
2008-09-07 16:34 <shapor> howso
2008-09-07 16:35 <flips> maxgid * maxuid * anymdode = huge
2008-09-07 16:36 <shapor> yeah but that is a vrey unlikely case
2008-09-07 16:36 <shapor> no reason to pay that max price for every file if you have only 10 or so on most systems
2008-09-07 16:36 <flips> how about a list post?
2008-09-07 16:36 <shapor> yeah, working on it ;)
2008-09-07 16:36 <shapor> recalling my one-liner from irc logs ;)
2008-09-07 16:37 <shapor> find / -xdev -type d -exec sh -c 'ls -l $1 | awk "/^\-/ {print \$1}" | sort -u |wc -l' {} {} \; | sort | uniq -c
2008-09-07 16:37 <shapor> heh
2008-09-07 16:37 <flips> another idea: record the actual user name in the fs, not just the id
2008-09-07 16:37 <flips> so that security will work when you copy a volume to a different system
2008-09-07 16:38 <flips> or at least some kind of mapping table
2008-09-07 16:38 <flips> which might well be an atom table
2008-09-07 16:39 <konrad> shapor: what's that bit of shell do?
2008-09-07 16:39 <flips> using the atom table we could map 32 bit uid and gid down to 16 or even 8 bits
2008-09-07 16:39 <shapor> its not quite what i wanted
2008-09-07 16:39 <shapor> but it basically breaks down the number of unique permissions sets per directory on the system
2008-09-07 16:40 <shapor> rather, it creates a distribution of uniqueness of permissions in a single directory
2008-09-07 16:41 <flips> shapor, is your tux3fs patch ready to go in, should I pull?
2008-09-07 16:42 <shapor> yeah
2008-09-07 16:44 <flips> whoops, I got 3 heads somehow
2008-09-07 16:45 <flips> damn, there they are
2008-09-07 16:46 <flips> sitting in your repo, I did view but didn't look all the way down
2008-09-07 16:46 <flips> now how do you revert a pull
2008-09-07 16:46 <flips> I guess this is a big flaming deficiency in hg
2008-09-07 16:47 <flips> ok, hg rollback works
2008-09-07 16:47 <flips> shapor, could you merge your heads before I pull?
2008-09-07 16:48 <shapor> how do i do that
2008-09-07 16:48 <flips> "hg merge"
2008-09-07 16:48 <flips> if it won't merge, tell it the version you want merged
2008-09-07 16:48 <flips> wish there was a simple delete head
2008-09-07 16:48 <shapor> abort: there is nothing to merge, just use 'hg update' or look at 'hg heads'
2008-09-07 16:49 <flips> right, you need to tell it exactly what head
2008-09-07 16:49 <flips> start with raponen's patch you added independently
2008-09-07 16:49 <flips> that might have conflicts
2008-09-07 16:49 <flips> because I adjusted a little
2008-09-07 16:49 <flips> hg is broken here
2008-09-07 16:51 <flips> http://www.selenic.com/pipermail/mercurial/2006-September/010628.html "Delete a branch/head"
2008-09-07 16:51 <shapor> hm ok
2008-09-07 16:51 <shapor> merged my old heads
2008-09-07 16:51 <shapor> now there is only one
2008-09-07 16:51 <shapor> i think all is well with my repo
2008-09-07 16:52 <flips> "added 12 changesets with 11 changes to 3 files"
2008-09-07 16:53 <flips> so I got all your braindumps with that pull too ;-)
2008-09-07 16:53 <flips> well they are now carved in the history of tux3 for eternity
2008-09-07 16:54 <flips> "You can use either:
2008-09-07 16:54 <flips> hg clone -r <head_you_want_to_keep> oldrepo newrepo
2008-09-07 16:54 <flips> (safest method)
2008-09-07 16:54 <flips> or hg strip <first cset of the branch you want to delete>
2008-09-07 16:54 <flips> (does in repo stripping, you need mq extension activated to have the
2008-09-07 16:54 <flips> strip command)"  http://www.selenic.com/pipermail/mercurial/2006-September/010628.html
2008-09-07 16:55 <flips> need to get in the habit of preparing a clean repo for pulling I think
2008-09-07 16:55 <flips> it can be automated
2008-09-07 16:55 <flips> or maybe In can tell pull to just pull the tip
2008-09-07 16:55 <konrad> any reason for choosing hg over git?
2008-09-07 16:56 <flips> it's _way_ nicer to use
2008-09-07 16:56 <flips> git has a bunch of fuzzy thinking that nobody fixed just because it came from linus
2008-09-07 16:56 <flips> it's about exactly as fast, in spite of being written in super slow python
2008-09-07 16:57 <konrad> I wasn't questioning the speed
2008-09-07 16:58 <flips> shapor, hg has pull -r, which I can use if I'm alert to see extra heads
2008-09-07 16:58 <flips> but I think the better approach is to clone the version you want me to pull, so I always pull from the same place, and you decide exactly what I pull
2008-09-07 16:58 <flips> knew you weren't
2008-09-07 16:58 <konrad> alright :)
2008-09-07 16:58 <flips> it's just amazing that mercurial is as fast as it is, in spite of being hobbled by python
2008-09-07 16:59 <flips> matt mackall - amazing hacker
2008-09-07 16:59 <konrad> heh
2008-09-07 17:00 <konrad> I'd bet it either employs a lot of the stuff python is fast at (because underneath it's in C) or it employs its own C extensions
2008-09-07 17:00 <konrad> python's dicts are really speedy, for example
2008-09-07 17:00 <flips> shapor, one thing I forgot to mention in my xattr post - we are going to use ext2 dirops for the atom table for the time being ;-)
2008-09-07 17:01 <flips> konrad, all of that
2008-09-07 17:01 <flips> plus it's a better design
2008-09-07 17:01 <flips> I bet it would kick git's tail if it were converted to c++
2008-09-07 17:01 <konrad> heh
2008-09-07 17:02 <flips> sk8 oclock
2008-09-07 17:02 <konrad> whoo!
2008-09-07 17:02 <konrad> enjoy
2008-09-07 17:02 <flips> we will
2008-09-07 17:02 <flips> going to meet shap for cocktails on the strand
2008-09-07 17:02 <flips> skating under the influence is fun and legal
2008-09-07 17:02 <flips> so far
2008-09-07 17:06 <konrad> heh
2008-09-07 19:24 <shapor> relatively legal anyway
2008-09-07 19:34 <konrad> this channel is logged :)
2008-09-07 19:48 <shapor> by me ;)
2008-09-07 20:07 <konrad> heh
2008-09-07 20:07 <konrad> me too, but it's not like I'm publicizing them
2008-09-07 20:07 <flips> konrad, completely legal
2008-09-07 20:07 <flips> unless of course you take out a baby buggy, just don't do that
2008-09-07 20:08 <flips> shapor, what say we put the logs up on tux3.org?
2008-09-07 20:09 <flips> and let googlebot chew on them
2008-09-07 20:09 <flips> help somebody learn something about skating maybe
2008-09-07 20:11 <flips> ok, howbig has to know how to include the size of variable sized items now
2008-09-07 20:12 <flips> I suppose easiest is to make the size of variable sized items a generic field
2008-09-07 20:13 <flips> something like [ kind:version:16,  atom:16, size:16, data[size] }
2008-09-07 20:14 <flips> or for a file data attribute: [ kind:version:16,  size:16, data[size] }
2008-09-07 20:14 <flips> hmm
2008-09-07 20:14 <flips> might be better to include the atom in the data
2008-09-07 20:14 <flips> { kind:version:16,  size:16,  atom:16 data[size - 2] }
2008-09-07 20:15 <flips> ok that seems better
2008-09-07 20:15 <flips> I'll make it so unless somebody has a better idea
2008-09-07 20:15 <flips> 0x1dea <- great magic number, got to use it for something
2008-09-07 20:51 <flips> "howmuch" makes its grand return
2008-09-07 20:52 <flips> howbig -> size of fixed attributes, howmuch -> size of variable attributes
2008-09-07 22:30 <shapor> flips: googlebot is already chewing on them
2008-09-07 22:31 <shapor> they are linked off shapor.com/tux3
2008-09-07 22:31 <flips> ah good
2008-09-07 22:31 <flips> lets see what comes up for tux3 sk8
2008-09-07 22:31 <flips> nothing
2008-09-07 22:31 <flips> wtf is up with googlebot
2008-09-07 22:31 <flips> bad googlebot
2008-09-07 22:50 <shapor> hm the lots are .txt, dunno if that has any effect
2008-09-07 23:13 <shapor> flips:  ah you pulled from me this time so i didn't have to merge my own changes, heh
2008-09-07 23:53 <flips> :-)