2008-08-18 00:01 <shapor> i thought i remember seeing that too
2008-08-18 00:13 <flips> [18233] main: fdsize64 failed for '(null)' (Bad file descriptor)
2008-08-18 00:14 <flips> better
2008-08-18 00:14 <flips> I'm still interested in the behavior when it tries to keep running anyway
2008-08-18 00:14 <flips> I'll just change the error() to warn()
2008-08-18 00:16 <flips> int fdsize64(int fd, uint64_t *size)
2008-08-18 00:16 <flips> {
2008-08-18 00:16 <flips> 	struct stat stat;
2008-08-18 00:16 <flips> 	if (fstat(fd, &stat))
2008-08-18 00:16 <flips> 		return -errno;
2008-08-18 00:16 <flips> 	if (S_ISREG(stat.st_mode)) {
2008-08-18 00:16 <flips> 		*size = stat.st_size;
2008-08-18 00:16 <flips> 		return 0;
2008-08-18 00:16 <flips> 	}
2008-08-18 00:16 <flips> 	return ioctl(fd, BLKGETSIZE64, size) ? -errno : 0;
2008-08-18 00:16 <flips> }
2008-08-18 00:16 <flips> maybe it should just return -1 like other libc stuff
2008-08-18 00:16 <flips> yes it should
2008-08-18 00:17 <flips> int fdsize64(int fd, uint64_t *size)
2008-08-18 00:17 <flips> {
2008-08-18 00:17 <flips> 	struct stat stat;
2008-08-18 00:17 <flips> 	if (fstat(fd, &stat))
2008-08-18 00:17 <flips> 		return -1;
2008-08-18 00:18 <flips> 	if (S_ISREG(stat.st_mode)) {
2008-08-18 00:18 <flips> 		*size = stat.st_size;
2008-08-18 00:18 <flips> 		return 0;
2008-08-18 00:18 <flips> 	}
2008-08-18 00:18 <flips> 	return ioctl(fd, BLKGETSIZE64, size);
2008-08-18 00:18 <flips> }
2008-08-18 00:29 <shapor> lgtm
2008-08-18 00:30 <shapor> flips: did you see my patch on the list
2008-08-18 00:30 <flips> not yet
2008-08-18 00:30 <shapor> makefile update + more 64 bit wanrings
2008-08-18 00:31 <flips> ah
2008-08-18 00:33 <flips> needs a little merge lovin
2008-08-18 00:34 <shapor> it wouldn't if you committed more often ;)
2008-08-18 00:34 <shapor> its been 8 hrs.. cmon, yer slackin
2008-08-18 00:35 <flips> you erred
2008-08-18 00:35 <flips> it's been 2 minutes
2008-08-18 00:35 <shapor> ah my cron job hasn't run
2008-08-18 00:35 <shapor> damn it
2008-08-18 00:35 <shapor> thats it, every minute
2008-08-18 00:37 <flips> um
2008-08-18 00:38 <flips> please, I am monitoring the http accesses
2008-08-18 00:38 <flips> you will dominate
2008-08-18 00:38 <flips> you now have 8, 1 minute apart
2008-08-18 00:39 <flips> there has to be an event driven way to do this
2008-08-18 00:39 <flips> I thought that was what rss was about
2008-08-18 00:41 <shapor> if your server can't take 1 pull per minute from me, that is sad
2008-08-18 00:41 <shapor> rss is bs
2008-08-18 00:41 <shapor> its just an xml formated page
2008-08-18 00:41 <flips> it can
2008-08-18 00:41 <shapor> ajax is supposedly the browser "push" technology
2008-08-18 00:41 <flips> it's my eyes when I scan the log
2008-08-18 00:42 <flips> I see all these pulls
2008-08-18 00:42 <shapor> but its really just javascript'ed pulls
2008-08-18 00:42 <shapor> of more xml
2008-08-18 00:42 <flips> lame
2008-08-18 00:42 <shapor> stupid heavy format
2008-08-18 00:42 <shapor> yep, polling is the way of the web
2008-08-18 00:43 <flips> hairy footed hippies on the steering committee methinks
2008-08-18 00:45 <flips> there, shapor.com seems to have mellowed a little
2008-08-18 00:45 <flips> how could I notify your pull if I wanted to, email?
2008-08-18 00:45 <flips> do some magic http access to your server?
2008-08-18 00:46 <shapor> i could just provide a repo you could push to
2008-08-18 00:46 <shapor> why bother with the overhead of requesting a pull ?
2008-08-18 00:46 <flips> then you poll it ;-)
2008-08-18 00:46 <flips> sure
2008-08-18 00:46 <flips> we should do that
2008-08-18 00:46 <shapor> well yeah, poll on the local box
2008-08-18 00:47 <flips> I want auto-push
2008-08-18 00:47 <flips> does hg do it?
2008-08-18 00:47 <shapor> doubtful
2008-08-18 00:47 <shapor> hrm perhaps
2008-08-18 00:47 <shapor> well you could poll locally too :P
2008-08-18 00:47 <flips> what I was thinking and making funny faces
2008-08-18 00:48 <flips> and I am supposed to fix inotify ;-)
2008-08-18 00:48 <shapor> heh
2008-08-18 00:48 <shapor> oh right.. for the kde guys?
2008-08-18 00:48 <shapor> it works well enough
2008-08-18 00:48 <flips> right
2008-08-18 00:48 <shapor> for this purpose
2008-08-18 00:48 <flips> they say it doesn't
2008-08-18 00:48 <flips> perhaps for this purpose
2008-08-18 00:49 <shapor> hrm an hg watcher that pushes upstream on local commits would be useful
2008-08-18 00:49 <shapor> could support git also
2008-08-18 00:49 <flips> let's get our whine in
2008-08-18 00:49 <shapor> why whine?
2008-08-18 00:50 <flips> you'll just code it I know
2008-08-18 00:50 <flips> no whining
2008-08-18 00:50 <flips> it would be good
2008-08-18 00:50 <flips> some way you just say source/target like zumastor
2008-08-18 00:50 <flips> and have replicated nets of source code
2008-08-18 00:50 <shapor> yeah certainly needs thought/first stab at it before whining
2008-08-18 00:51 <shapor> bitbucket/github guys would be interested probably
2008-08-18 00:51 <flips> well, starting to get late
2008-08-18 00:51 <shapor> i wonder if they have any programs to do it already
2008-08-18 00:51 <shapor> not even 1 yet
2008-08-18 00:51 <flips> probably matt is no dummy
2008-08-18 00:52 <flips> I was just in the middle of setting up some btree unit testing
2008-08-18 00:52 <flips> got tired of testing btree in the live app
2008-08-18 00:52 <flips> things like advance should be tested in isolation
2008-08-18 00:52 <flips> and lots of other things
2008-08-18 00:52 <flips> if btrees are ever expected to be solid
2008-08-18 00:58 <shapor> ok i cleaned up the 64 bit warning on the most recent code ;)
2008-08-18 00:58 <flips> patch?
2008-08-18 00:58 <shapor> mailed
2008-08-18 00:58 <shapor> this is already getting old
2008-08-18 00:58 <flips> right, I should pull from you
2008-08-18 00:59 <shapor> i need to learn how to make a tree public with hg
2008-08-18 00:59 <shapor> tomorrow
2008-08-18 01:00 <flips> I'll just look in my httpd.conf when you're ready
2008-08-18 01:01 <flips> basically nothing to do
2008-08-18 01:01 <flips> just give access to it
2008-08-18 01:02 <flips> if I can see the repo directory I can pull
2008-08-18 01:02 <flips> and unlike git this is efficient
2008-08-18 01:03 <shapor> ah cool
2008-08-18 01:04 <shapor> that is easy for me to do, just create a symlink
2008-08-18 01:04 <flips> yes
2008-08-18 01:04 <shapor> the .hg dir?
2008-08-18 01:04 <shapor> or the root of my checkout?
2008-08-18 01:04 <flips> the root I think
2008-08-18 01:04 <flips> let me check
2008-08-18 01:05 <flips> the root of your repo, not the .hg
2008-08-18 01:05 <shapor> http://shapor.com/tux3/hg/
2008-08-18 01:06 <flips> we shoulda used your last patch to test
2008-08-18 01:06 <shapor> i canlt clone from it though
2008-08-18 01:06 <shapor> are you sure its that simple?
2008-08-18 01:06 <shapor> can't*
2008-08-18 01:06 <flips> yes
2008-08-18 01:07 <flips> you should be able to clone
2008-08-18 01:07 <shapor> abort: 'http://shapor.com/tux3/hg/' does not appear to be an hg repository!
2008-08-18 01:07 <flips> does it have a .hg?
2008-08-18 01:07 <shapor> yes
2008-08-18 01:07 <shapor> http://shapor.com/tux3/hg/.hg/
2008-08-18 01:07 <flips> just a sec
2008-08-18 01:09 <shapor> i dont think its as simple as you say it is
2008-08-18 01:10 <flips> ah, you can clone like that
2008-08-18 01:10 <shapor> ah
2008-08-18 01:10 <flips> but not pull
2008-08-18 01:10 <shapor> yeah 
2008-08-18 01:11 <shapor> er are you sure?
2008-08-18 01:11 <shapor> i think you can
2008-08-18 01:11 <flips> ah
2008-08-18 01:11 <shapor> static-http:
2008-08-18 01:11 <flips> you can pull after you clone ;-)
2008-08-18 01:11 <flips> no you can't
2008-08-18 01:11 <shapor> http://www.selenic.com/mercurial/wiki/index.cgi/StaticHTTP
2008-08-18 01:11 <flips> clone is happy to clone, but pull says no repo, bad message
2008-08-18 01:12 <shapor> are you using static-http ?
2008-08-18 01:12 <flips> static wha?
2008-08-18 01:12 <shapor> hg pull
2008-08-18 01:12 <shapor> pulling from static-http://shapor.com/tux3/hg
2008-08-18 01:13 <shapor> you can't just use the http: prefix
2008-08-18 01:13 <shapor> that expect a mercurial cgi script on the other end
2008-08-18 01:13 <shapor> static-http: prefix expects just the regular old files
2008-08-18 01:13 <shapor> on the other end
2008-08-18 01:13 <shapor> see the link i sent above
2008-08-18 01:13 <flips> ok, here goes
2008-08-18 01:14 <flips> hg pull static-http://shapor.com/tux3/hg
2008-08-18 01:14 <flips> abort: no repo found!
2008-08-18 01:16 <shapor> you have to be in an hg tree
2008-08-18 01:16 <flips> hg clone static-http://shapor.com/tux3/hg <- works
2008-08-18 01:16 <shapor> run that pull command inside your tux3 dir
2008-08-18 01:16 <flips> yes that works
2008-08-18 01:16 <flips> now... it shows as directory hg
2008-08-18 01:17 <flips> that is not too good style
2008-08-18 01:17 <shapor> doh
2008-08-18 01:17 <flips> would be better named shapor
2008-08-18 01:17 <flips> well tux3
2008-08-18 01:17 <flips> better
2008-08-18 01:17 <shapor> tux3-shapor?
2008-08-18 01:17 <flips> sure
2008-08-18 01:18 <shapor> you can call it whatever you like though
2008-08-18 01:18 <flips> http pull this way is efficient enough
2008-08-18 01:18 <shapor> i think
2008-08-18 01:18 <flips> way better than email
2008-08-18 01:18 <shapor> hell yeah
2008-08-18 01:18 <flips> calling it hg is lame ;-)
2008-08-18 01:19 <shapor> makes sense tux3/hg
2008-08-18 01:19 <shapor> its the mercurial repo for tux3 ;)
2008-08-18 01:19 <shapor> renamed to shapor-tux3
2008-08-18 01:21 <shapor> ok i commited a change
2008-08-18 01:21 <shapor> pull from me
2008-08-18 01:22 <flips> dinner time here
2008-08-18 01:22 <flips> ok
2008-08-18 01:22 <flips> hg pull static-http://shapor.com/tux3-shapor
2008-08-18 01:22 <flips> abort: HTTP Error 403: Forbidden
2008-08-18 01:23 <shapor> hg pull static-http://shapor.com/tux3/shapor-tux3
2008-08-18 01:25 <flips> worked
2008-08-18 01:25 <flips> now I need to see the diff
2008-08-18 01:26 <shapor> oh cool, hg has support for all kinds of hooks
2008-08-18 01:26 <shapor> look in man hgrc
2008-08-18 01:26 <shapor> under "hooks"
2008-08-18 01:27 <flips> yes, what an excellent use of time
2008-08-18 01:27 <flips> ok, I need to pay attention the the family now
2008-08-18 01:27 <shapor> catch ya tomorrow
2008-08-18 02:50 -!- pgquiles(~pgquiles@172.Red-83-38-37.dynamicIP.rima-tde.net) has joined #tux3
2008-08-18 10:11 -!- pgquiles(~pgquiles@251.Red-81-37-107.dynamicIP.rima-tde.net) has joined #tux3
2008-08-18 12:28 <boom> So... is tux3 going to really be a "ZFS killer?"
2008-08-18 14:23 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #tux3
2008-08-18 14:28 <flips> boom, zfs is self killing imho
2008-08-18 14:28 <flips> "zfs is a rampant layer violation" -- akpm
2008-08-18 14:28 <flips> tux3 + lvm3 will cover the checkbox items of zfs
2008-08-18 14:29 <flips> not that I am wildly excited about the idea of checksumming metadata, but tux3 will eventually have that too, as an option
2008-08-18 14:30 <flips> tux3 will have the immense advantage of running on Linux
2008-08-18 14:30 <flips> as long as Sun keeps being idiotic about the zfs license, zfs will not
2008-08-18 15:03 <boom> flips: The licensing was something that had me sort of confused. Is Linux's policy to never include non-GPL'd code in the kernel itself?
2008-08-18 15:13 <flips> more specific than that: the code has to be GPL v2
2008-08-18 15:13 <flips> v3 will no do
2008-08-18 15:13 <boom> Really..?
2008-08-18 15:13 <boom> Whose decision was that, Linus'?
2008-08-18 15:13 <flips> really
2008-08-18 15:14 <flips> Linus says he can't change is mind because every copyright holder would have to agree
2008-08-18 15:14 <flips> there are hundreds, some of them even died
2008-08-18 15:14 <boom> Okay, so even still, if the code is available (as it is for freebsd) what's preventing someone from just building and loading it as a third party module?
2008-08-18 15:15 <flips> you can, but binary modules without proper license are in a legal gray zone
2008-08-18 15:15 <flips> the code will certainly not go into mainlin until Sun adds a GPL v2 license
2008-08-18 15:15 <boom> Isn't that "gray zone" what's currently being filled by things like nvidia's binary driver?
2008-08-18 15:16 <flips> nvidia's driver has caused all kinds of problems
2008-08-18 15:16 <flips> being occupied by is a better term than being filled by
2008-08-18 15:16 <flips> filled sounds like satisfies
2008-08-18 15:16 <boom> That's true
2008-08-18 15:17 <boom> And would tux3 be GPL2, then?
2008-08-18 15:17 <flips> user space code is gpl v3, kernel code is gpl v2
2008-08-18 15:17 <boom> Okay, thanks for the info.
2008-08-18 15:18 <boom> I've gotta say, though, I've developed a growing respect for BSD-licensed code
2008-08-18 15:18 <flips> say, that reminds me, it is about time to collect my beer from Eben
2008-08-18 15:18 <flips> bsd is awesome
2008-08-18 15:18 <boom> I'm sure that would open you guys up to all kinds of abuse from the rest of capitalism though
2008-08-18 15:18 <flips> what would?
2008-08-18 15:18 <flips> oh
2008-08-18 15:18 <flips> bsd
2008-08-18 15:18 <flips> yes, I am not in this to give code to msoft
2008-08-18 15:19 <boom> I can support that.
2008-08-18 15:19 <boom> Hah.
2008-08-18 15:20 <boom> I wasted six hours today on their asses.
2008-08-18 15:21 <boom> Do you guys have a projected timeline?
2008-08-18 15:23 <flips> a draft roadmap is on the mailing list, revised one will go up later today
2008-08-18 15:25 <boom> Ah, I suppose I should subscribe to that bad boy.
2008-08-18 15:25 <flips> ;-)
2008-08-18 15:46 <flips> welcome to tux3
2008-08-18 15:51 <boom> Many thanks.
2008-08-18 15:51 <flips> np
2008-08-18 16:00 <flips> g99 -g -Wall buffer.c diskio.c btree.c && ./a.out foodev
2008-08-18 16:00 <flips> root at 0
2008-08-18 16:00 <flips> leaf at 1
2008-08-18 16:00 <flips> btree leaf with 0 entries
2008-08-18 16:00 <flips> leaf free = 3c
2008-08-18 16:00 <flips> btree unit tests starting to happen
2008-08-18 16:00 <flips> little leaves to make bushy trees
2008-08-18 17:25 <bh> folks ;0
2008-08-18 17:25 <bh> :)
2008-08-18 17:25 <bh> 
ACTION finds that the backlog has been truncated

2008-08-18 17:44 -!- konrad(~konrad@c-24-16-74-109.hsd1.wa.comcast.net) has joined #tux3
2008-08-18 22:48 <flips> hey shapor
2008-08-18 22:48 <flips> the btree unit test produces buggy results
2008-08-18 22:48 <flips> interested?
2008-08-18 22:48 <flips> seems I broke it on the port from ddsnap
2008-08-18 22:49 <flips> bh, you're coming up for air
2008-08-18 22:49 <flips> ?
2008-08-18 23:02 <bh> eh ?
2008-08-18 23:02 <bh> what's air ?
2008-08-18 23:03 <bh> just working late at night as usual
2008-08-18 23:54 <shapor> flips: ah i see
2008-08-18 23:55 <shapor> the insert after the split inserts to the wrong one
2008-08-18 23:55 <flips> right
2008-08-18 23:55 <flips> know why yet?
2008-08-18 23:55 <shapor> didn't look at the code
2008-08-18 23:55 <flips> I was just going in there
2008-08-18 23:56 <shapor> i'm guessing similar to the fleaf, er whatever its called now, dleaf?
2008-08-18 23:56 <shapor> it was broken in the same way initially i think
2008-08-18 23:56 <flips> of course it works in ddsnap, I broke it :-/
2008-08-18 23:56 <flips> unit tests noticed