| 00:04.59 | sunrunner20 | I need to build a linux install from scratch so I understand everything |
| 00:10.14 | *** join/#edev risca (~risca@wi-secure-434.cc.umanitoba.ca) |
| 00:11.39 | SpeedEvil | sunrunner20: Collisions don't matter, if they are vanishingly unlikely. |
| 00:11.45 | SpeedEvil | sunrunner20: And unengineerable. |
| 00:14.01 | MonMotha | there are several well known cryptosystems that have proven reasonably secure in the face of widespread attack |
| 00:14.22 | SpeedEvil | For hashes, you can somewhat cheat. |
| 00:14.36 | SpeedEvil | Simply take all the hashes you can think of, hash, and xor the result. |
| 00:15.01 | SpeedEvil | It's truly spectacularly unlikely that by xoring four hashes you solve the fifth. |
| 00:15.05 | MonMotha | In fact, to my knowledge, BD+ remains unbroken, and that's a HUGE target. Obviously people have attacked it by simply recovering the deciphered stream from "legitimately authorized" decryptors, but I don't believe the crypto itself has been sucessfully attacked. |
| 00:15.21 | MonMotha | I'm not sure why XOR is particularly relevant |
| 00:15.48 | SpeedEvil | MonMotha: Simply to add the values of various hashes together |
| 00:16.08 | SpeedEvil | It of course adds no strength on its own |
| 00:16.09 | MonMotha | but yes, a good hash has a likelihood of collision that's essentially negligible for random/real-world inputs and has no way to deliberately cause a collision |
| 00:16.20 | MonMotha | the SHA-2 family still meets both criteria, though there are some worries |
| 00:16.36 | MonMotha | SHA-1 and MD5 are of course totally broken in the latter regard (and MD5 is a bit small to even meet the former) |
| 00:17.06 | MonMotha | IIRC, SHA-3 is still in contest phase, but some of the contenders look good |
| 00:17.35 | MonMotha | then again, "I Am Not A Cryptographer", though I trust myself to engineer an existing cryptosystem into a new application without making too many mistakes (I of course ask for peer review) |
| 00:18.11 | MonMotha | I'd never even attempt to invent my own cipher or hash algo, and I stay away from rolling my own crypto session systems (ala SSL/TLS) as much as possible in preference to using known implementations |
| 00:18.21 | MonMotha | since it's so easy to make a silly mistake that totally breaks things |
| 00:18.51 | SpeedEvil | Key managment and not reusing stuff isn't stuff you want to make silly mistakes on. |
| 00:19.08 | SpeedEvil | ps3/xbox |
| 00:19.28 | MonMotha | indeed |
| 00:19.38 | MonMotha | a nonce really does mean a number used *ONCE* |
| 00:19.57 | MonMotha | key "management" isn't generally too hard. Key *distribution* on the other hand... |
| 00:20.08 | MonMotha | and there are still some subtleties in key "management" that'll get you |
| 00:20.54 | SpeedEvil | 'hmm - that seems secure' - from a programmers POV = maths - but there was a neat theorem that broke that in 1890. |
| 00:20.54 | MonMotha | I also remember being scared as heck of counter mode, though I must admit it seems like it's actually a fairly secure mode that's tough to get wrong |
| 00:20.56 | MonMotha | unlike, say, CBC |
| 00:22.12 | MonMotha | reading about the failures of WEP almost makes me laugh at this point |
| 00:22.17 | MonMotha | I'm a bit sad that the IEEE came up with that |
| 00:24.03 | SpeedEvil | It is true. It's wired equivalent privacy. |
| 00:24.11 | SpeedEvil | If the wire is dangling into a dark alley |
| 00:24.57 | sunrunner20 | oh SpeedEvil, that makes me sad bash.org is dead |
| 00:25.11 | MonMotha | the joke I heard was that they wanted wired *equivalent* privacy, not "good security", so they deliberately came up with something bad to highlight that wired networks are not inherently secure :) |
| 00:25.20 | SpeedEvil | :) |
| 00:25.44 | MonMotha | there's always qdb |
| 00:25.57 | MonMotha | that said, WPA2 seems to have held up well |
| 00:25.57 | sunrunner20 | or reddit >_> |
| 00:26.12 | MonMotha | and, heck, TKIP did pretty well given what they had to work with |
| 00:26.18 | *** join/#edev NiveusLuna (~niveuslun@fedora/NiveusLuna) |
| 00:26.32 | MonMotha | I'm also not aware of any attacks on the 802.15.4 security layer, though it's a much smaller target |
| 00:27.49 | SpeedEvil | I'm kinda waiting for the UK to get our smart-meters installed only to find out they are hideously vulnerable in some way. |
| 00:28.32 | SpeedEvil | Which reminds me, I need to find if there is an appropriate person/committee to poke to leave open the option of net-metering. |
| 00:28.57 | SpeedEvil | (to ensure the smart meters can read 'backwards' power. |
| 00:29.20 | MonMotha | heh, the smart meters in the USA are often pretty pathetic |
| 00:29.43 | MonMotha | most of the IR ones are TOTALLY unsecured. Like anybody with line of sight, an IR emitter out of a remote control, and an Arduino can turn off your power |
| 00:29.55 | sunrunner20 | lol |
| 00:29.56 | MonMotha | the 802.15.4 ones are *sometimes* a bit better |
| 00:30.47 | SpeedEvil | MonMotha: I've got a power meter that logs at 6s intervals. |
| 00:31.01 | SpeedEvil | It's truly remarkable how much you can extract from this, with a little thought. |
| 00:31.03 | MonMotha | some of them are even more frequent than that |
| 00:31.06 | landley | g1powermac: all working encryption depends on some secret. If you feed somebody the key to play back the data, it's all over. |
| 00:31.09 | MonMotha | and yes, they can put some amazing stats |
| 00:31.35 | MonMotha | *pull |
| 00:31.45 | SpeedEvil | Everything from fridge openings, time you have breakfast, how many cups of coffee you have a day, indoor temperature, how long you shower, ... |
| 00:31.51 | SpeedEvil | TV program you're watching. |
| 00:32.01 | MonMotha | the latter is a bit hard :) |
| 00:32.01 | g1powermac | landley, that is true, which is why it would be interesting if you could manipulate the meaning of the data so no encryption would be needed |
| 00:32.12 | SpeedEvil | MonMotha: It's a bit hard. It's not a lot hard though. |
| 00:32.38 | SpeedEvil | MonMotha: If you have ~10W or so delta in power as the signal changes - you can correlate quite well - given ~1800 samples. |
| 00:32.40 | MonMotha | with 6s granularity? I'd think it would only work very well if you limited it to a small subset of available programming |
| 00:32.49 | landley | steganography: the art of hiding your data in a stegasaurus. |
| 00:33.09 | landley | Encoding missing and present spines on the back, I guess? |
| 00:52.07 | *** join/#edev djerome (~djerome@ip68-2-20-108.ph.ph.cox.net) |
| 01:11.55 | *** join/#edev risca (~risca@wi-secure-434.cc.umanitoba.ca) |
| 02:04.22 | *** join/#edev factor (~factor@r74-195-184-248.msk1cmtc01.mskgok.ok.dh.suddenlink.net) |
| 03:01.42 | *** join/#edev risca (~risca@wnpgmb0903w-ds01-177-34.dynamic.mtsallstream.net) |
| 03:18.38 | *** join/#edev prpplague (~prpplague@ppp-70-249-149-70.dsl.rcsntx.swbell.net) |
| 03:33.51 | *** join/#edev gandhijee_ (~akp@50.12.169.99) |
| 04:56.55 | *** join/#edev gandhijee_ (~akp@50.12.169.99) |
| 05:25.43 | *** mode/#edev [+o prpplague] by ChanServ |
| 05:53.50 | *** join/#edev clio (~andrej@85.159.109.222) |
| 06:28.21 | *** join/#edev War2 (~war2@unaffiliated/war2) |
| 07:13.54 | *** join/#edev rsFF (~rsFF@2.80.53.127) |
| 07:16.12 | *** join/#edev slashbeast (~slashbeas@funtoo/core/slashbeast) |
| 08:20.30 | *** join/#edev factor (~factor@r74-195-184-248.msk1cmtc01.mskgok.ok.dh.suddenlink.net) |
| 08:26.55 | *** join/#edev icarusfactor (~factor@r74-195-184-248.msk1cmtc01.mskgok.ok.dh.suddenlink.net) |
| 08:33.29 | *** join/#edev gandhijee_ (~akp@50.12.169.99) |
| 09:30.30 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 10:38.36 | *** join/#edev gandhijee_ (~akp@50.12.169.99) |
| 10:43.25 | *** join/#edev icanicant (~klawson@213.218.221.154) |
| 10:58.59 | *** join/#edev g1powermac (~g1powerma@rrcs-173-197-150-24.west.biz.rr.com) |
| 10:58.59 | *** join/#edev g1powermac (~g1powerma@unaffiliated/g1powermac) |
| 11:37.22 | *** join/#edev lfelipe (~lfelipe@201.82.19.44) |
| 11:37.22 | *** join/#edev lfelipe (~lfelipe@enlightenment/developer/lfelipe) |
| 11:54.43 | *** join/#edev War2 (~war2@230.Red-217-125-111.staticIP.rima-tde.net) |
| 11:54.43 | *** join/#edev War2 (~war2@unaffiliated/war2) |
| 13:31.25 | slashbeast | Anyone know where is Landley? |
| 13:37.01 | Mirell | Probably asleep? |
| 13:37.06 | Mirell | Or on his way to work? |
| 13:37.54 | av500 | or goofing off |
| 13:38.09 | *** join/#edev gandhijee_ (~akp@50.12.169.99) |
| 13:38.53 | NiveusLuna | If anyone has the time, I'd appreciate some feedback on my HTTPS Everywhere 2.0 question before I package it. |
| 13:41.33 | NiveusLuna | On the mailing list, not in here. |
| 13:42.00 | NiveusLuna | .... |
| 13:42.02 | NiveusLuna | whoops |
| 13:42.06 | NiveusLuna | this is #edev, not #epel |
| 13:42.10 | NiveusLuna | I feel stupid now. XD |
| 13:47.42 | Mirell | Heh |
| 13:58.51 | NiveusLuna | That explains why nirik isn't in here, doesn't it? :P |
| 14:03.30 | *** join/#edev FrankSansC (~quassel@LVelizy-156-45-9-168.w80-11.abo.wanadoo.fr) |
| 14:14.07 | *** join/#edev JoeLlama (~snork@unaffiliated/joellama) |
| 14:49.16 | *** join/#edev sunrunner20 (~jamesdoe@pool-71-123-194-101.dllstx.fios.verizon.net) |
| 15:08.40 | *** join/#edev GPSFan (~kenm@64.92.145.112) |
| 15:08.41 | *** mode/#edev [+o GPSFan] by ChanServ |
| 15:12.38 | slashbeast | Mirell: I didnt saw him around for a few days. |
| 15:15.20 | Mirell | slashbeast: He was here yesterday |
| 15:23.37 | sunrunner20 | wtf, i'm not giving livestrong my facebook info |
| 15:55.51 | *** join/#edev bgamari (~ben@physicsnat56.physics.umass.edu) |
| 15:56.46 | *** join/#edev prpplague (~danders@192.94.92.14) |
| 16:04.38 | NiveusLuna | doesn't have any facebook info to give. |
| 16:14.00 | *** join/#edev ccole (~cole@coledd.com) |
| 16:20.56 | *** join/#edev landley (~landley@140.242.26.2) |
| 16:34.47 | *** join/#edev sndcrb (~sndcrb@dsl-hkibrasgw2-fef8de00-126.dhcp.inet.fi) |
| 16:35.52 | landley | And lo, it was morning. In accordance with the prophecy. |
| 16:36.09 | av500 | alas |
| 16:38.46 | Mirell | Alas? |
| 16:39.44 | av500 | http://en.wikipedia.org/wiki/Alas |
| 16:39.54 | av500 | 1st one |
| 16:40.22 | Mirell | Yes, thank you, I know what the word means. |
| 16:42.11 | av500 | ah ,sorry |
| 16:42.59 | av500 | should have said "so, it has come to this" |
| 16:57.36 | landley | Well it _has_... |
| 16:59.30 | av500 | landley: btw, does that look familiar? http://www.flickr.com/photos/av500/4807347293/in/set-72157624533739048 |
| 17:49.41 | *** join/#edev risca (~risca@wi-secure-4481.cc.umanitoba.ca) |
| 18:00.40 | landley | av500: I think I gave it to mirell. :) |
| 18:01.31 | landley | (Actually, that might not be a tuxphone. Although a later implementation of the same thing would make even _less_ sense...) |
| 18:01.46 | av500 | landley: ? |
| 18:03.48 | calculus | is it a hexagon evm? |
| 18:04.01 | av500 | snapdragon evm |
| 18:04.19 | av500 | with allegedly insanely high price |
| 18:15.37 | landley | Oh, the ancient ones in the lab! |
| 18:15.44 | landley | Right, those were obsolete and we never used them. |
| 18:15.58 | landley | There were like 3 of them still hooked up in a corner in case support calls ever came in, but they predated comet boards. |
| 18:16.11 | landley | They were QDSP6v1 and we were already on v2 (which is what shipped in the Nexus One). |
| 18:16.22 | landley | v3 had taped out but wasn't shipping yet, and v4 was in development. |
| 18:16.37 | landley | never saw one with a phone handset attached, but *shrug* |
| 18:17.14 | landley | Basically, that hardware predated my time at qualcomm. :) |
| 18:17.19 | av500 | :) |
| 18:19.54 | landley | Ok, I admit I'm reading lwn.net a week behind everybody else until I get around to renewing my subscription, but I keep seeing posts and going "I blogged about that over a year ago"... |
| 18:22.13 | landley | I mean, http://lwn.net/Articles/482557 has me going "oh, is it August again?" because it seems somewhere between http://landley.net/notes-2010.html#13-08-2010 and http://landley.net/notes-2011.html#10-08-2011 to me... |
| 18:22.47 | Mirell | landley: Your data model for databases (recent blog post) fails when a company is adding 1TB of data a day. |
| 18:22.57 | landley | Mirell: oh sure. |
| 18:23.12 | landley | there are high end customers that need something very specific. |
| 18:23.22 | landley | However, I'm not so sure Oracle's a good choice there either. That's data warehouse territory. :) |
| 18:23.33 | Mirell | landley: They're using Riak |
| 18:23.36 | landley | You're looking off the north end of mapreduce there. |
| 18:24.15 | landley | I was describing a disruptive technology that started by eating 10% of Oracle's market and has now passed 50% of it with 90% in sight. |
| 18:24.41 | Mirell | Yah, we're trying to force Oracle out. ^^ |
| 18:25.10 | landley | Oracle is DEC. You're the "big iron pushing down" and the in-memory stuff is the PC gnawing upwards. |
| 18:25.29 | landley | They're between a rock and a pointy bit. |
| 18:26.02 | landley | I'm serious about failed business models going down in a blaze of IP litigation. |
| 18:26.26 | landley | "We can't make you use our stuff but we can make you pay for it anyway!" Not a strategy with legs, that... |
| 18:26.28 | Mirell | Oh, landley, one nice thing about ZFS, L2ARC on SSDs while still having your data on slower but bigger HDDS |
| 18:26.50 | landley | Could you unpack that sentence slightly? |
| 18:26.53 | landley | what's l2ARC? |
| 18:27.05 | av500 | lharc? |
| 18:27.16 | Mirell | Layer 2 ARC, disk cache basically |
| 18:27.21 | landley | Ah. |
| 18:27.38 | Mirell | ARC is the main disk cache |
| 18:27.41 | Mirell | Which resides in RAM |
| 18:27.42 | landley | The next step beyond the "separate journal" stuff the ext3 guys were doing a decade ago. |
| 18:28.09 | landley | I thought the Linux guys were trying to implement this as a _seperate_ filesystem, and it kept getting caught up in the endless unionfs wars... |
| 18:28.26 | landley | (Really, unionfs needs to be something the vfs does, like ramfs.) |
| 18:28.45 | landley | Alas, with Val Aurora retired, progress has stopped again... |
| 18:29.11 | Mirell | It's nice because you can stick a 512GB SSD in your system, and have 10,000 RPM 2TB platter disks, and nicely optimize things |
| 18:29.40 | landley | Cool. Last I heard the btrfs guys were doing something like that, but I generally wait for stuff to be finished before worrying about it too much... |
| 18:30.03 | landley | (And there was some argument about using the container infrastructure for some of this...) |
| 18:30.07 | landley | needs to catch up on linux-kernel. |
| 18:30.13 | landley | I'm about... (checks) 3 years behind. |
| 18:30.17 | Mirell | Yah, Brtfs doesn't seem to have RAID yet either |
| 18:30.26 | landley | we've got a RAID layer? |
| 18:30.36 | landley | is pretty sure that Linux had one of those. |
| 18:30.39 | landley | Circa 2.2 or so... |
| 18:30.42 | Mirell | True, there's a layer. |
| 18:31.01 | Mirell | ZFS just handles that for you. |
| 18:31.08 | landley | Why? |
| 18:31.31 | Mirell | So you can tag disks as hot backups, establish RAID pools, et cetera |
| 18:32.03 | landley | goes to check the dm manual. Pretty sure there were knobs for this... |
| 18:32.13 | landley | I last had to care when people were threatening to add this stuff to busybox in 2006. |
| 18:33.21 | landley | I admit that if zfs cured cancer I still wouldn't care until there was a linux-compatible clone of it under a non-stupid license that wasn't owned by a patent troll. |
| 18:33.33 | landley | But I still haven't quite seen what the fuss is about... |
| 18:33.56 | landley | "It does things these other tools already do". That's nice... |
| 18:34.11 | landley | I suppose I'd really need to be a sysadmin of large data stores to have context for being excited here. |
| 18:34.41 | SpeedEvil | What would be really nice would be a sane filesystem and swap subsystem that knew how to deal with fake block devices which are really flash. |
| 18:34.53 | landley | It's like those SAS enclosures Dell was so excited about being able to sell, because the margins were so enormous, even after they gilded (ahem, "gold plated") the connectors... |
| 18:35.16 | landley | SpeedEvil: I take it Thomas Gleixner's mtd layer didn't do it for you? |
| 18:35.17 | SpeedEvil | Live defragmetation during write being almost free, for example. |
| 18:35.33 | SpeedEvil | landley: I mean for SD cards. |
| 18:35.36 | SpeedEvil | landley: And emmc. |
| 18:35.58 | SpeedEvil | Which are really flash, but present like a block device, but you have to know they're flash if you want more than 3k/second from them. |
| 18:36.01 | landley | And most of the flash I've seen has some pretty nasty erase/write latencies. It's just reads that are "almost free"... |
| 18:36.03 | SpeedEvil | I'm not familiar with that. |
| 18:36.06 | SpeedEvil | Indeed. |
| 18:36.15 | kergoth | there was a presentation at elc about performance optimization for those devices.. didn't catch that one though |
| 18:37.17 | SpeedEvil | I mean that if you want to write a block of 100K of data, it's essentially as cheap to read another 30K of data from somewhere else which is fragmented, and write 130K, to make it contiguous, and avoid later rewrite penalties. |
| 18:37.32 | SpeedEvil | This _really_ hurts with swap. |
| 18:37.50 | landley | SpeedEvil: http://www.linux-mtd.infradead.org/ |
| 18:38.00 | SpeedEvil | hmm - I was failing google |
| 18:38.34 | SpeedEvil | That's not quite what I'm meaning. |
| 18:39.04 | SpeedEvil | That's - as I understand it - intended for dealing with actually raw flash, not stuff like eMMC and SD, which pretends to be a perfect block device. |
| 18:39.24 | SpeedEvil | Actually raw flash in many ways is an easier problem. |
| 18:40.22 | landley | Apparently mtdblock is deprecated in favor of ubi: http://www.linux-mtd.infradead.org/doc/ubi.html |
| 18:40.51 | landley | (Although at work they're still just using mtdblock and putting ext3 on top of it, which is sort of horrifying, but oh well...) |
| 18:40.56 | SpeedEvil | Raw flash seems to be going away. |
| 18:41.27 | landley | SpeedEvil: not from the boards I'm using. |
| 18:41.33 | c1de0x | hey landley |
| 18:41.38 | SpeedEvil | Well - I mean from mobile phone space and stuff. |
| 18:41.42 | landley | hey c1de0x |
| 18:41.45 | SpeedEvil | Not 'small' ones at the moment. |
| 18:41.46 | c1de0x | did you ever have a chance to look at my distcc issues? |
| 18:42.29 | landley | I was down with cedar pollen for a couple days this week. (Antihistamines knock me out, slept 14 hours tuesday night.) |
| 18:42.46 | landley | I'm just now catching up, but I'm doing the toybox release before getting back to aboriginal. |
| 18:42.56 | landley | I hope to get all caught up this weekend. |
| 18:43.52 | landley | SpeedEvil: the problem with non-raw flash is... heh. |
| 18:44.10 | landley | Always a "what?" moment when a quick google for a topic turns up somebody quoting somethign I wrote. :) |
| 18:44.11 | landley | http://forum.doozan.com/read.php?2,3365,3365,quote=1 |
| 18:44.20 | landley | Anyway, collected links and spiel on the topic. |
| 18:44.42 | landley | c1de0x: Saturday, I should be able to give your distcc issue the time it deserves. |
| 18:44.52 | landley | Sorry, I was at work until 8pm last night. Crazy week... |
| 18:48.27 | c1de0x | <PROTECTED> |
| 18:48.46 | c1de0x | firstly. gotta say i really love the work you've done on aboriginal. |
| 18:48.55 | c1de0x | no worries. |
| 18:48.59 | c1de0x | i fu |
| 18:49.07 | c1de0x | i figured as much. |
| 18:49.46 | c1de0x | i've built a couple of build-control-images... and now use it to do automated weekly builds of tools ;) |
| 18:50.22 | c1de0x | its a lot better than the scripts i had hacked up to do things. |
| 18:50.22 | c1de0x | and the ability to just drop into a qemu shell is fantastic ;) |
| 19:00.28 | landley | c1de0x: glad you like it. :) |
| 19:00.42 | c1de0x | i really do. |
| 19:00.42 | landley | I've been pondering just writing my own distcc replacement (based on ccwrap) for a while now. |
| 19:00.55 | landley | The existing distcc is very nice, but not _quite_ designed to do what I want it to. |
| 19:01.14 | landley | And when it doesn't work, it silently falls back to local builds... |
| 19:01.23 | c1de0x | landley: well. i can't promise to help write, but i can promise to test and submit patches/bugs ;) |
| 19:01.28 | landley | Woot. |
| 19:01.40 | c1de0x | i o |
| 19:01.58 | landley | Toybox is taking up about 80% of my spare coding time at the moment, but aboriginal is the next higest priority. |
| 19:02.04 | landley | (Ok, it goes day job, toybox, aboriginal.) |
| 19:02.07 | c1de0x | i often need to compile stuff for embedded targets... and even on my 24-core devil-machine... qemu is a bitch |
| 19:02.13 | landley | but the day job has finite hours. In theory... |
| 19:02.17 | c1de0x | hehe. |
| 19:02.35 | landley | Did you grind through the giant 260 page presentation? |
| 19:02.36 | prpplague | landley: only in theory |
| 19:02.36 | av500 | landley: the netra behaving now? |
| 19:02.54 | landley | The "future optimization" section is still (sadly) mostly relevant. |
| 19:03.00 | c1de0x | yeah. is toybox significantly smaller (bin-size) than busybox? |
| 19:03.00 | c1de0x | (for equivalent applets) |
| 19:03.08 | landley | Chop out tinycc's preprocessor, it's about twice as fast. |
| 19:03.19 | landley | Get gigabit ethernet working on all the boards (it mostly doesn't off of x86). |
| 19:03.31 | landley | c1de0x: dunno, haven't checked. |
| 19:03.32 | c1de0x | landley: which presentation? |
| 19:03.43 | landley | I'm going for _simpler_. Small and fast are mostly side effects of that. |
| 19:03.49 | c1de0x | ok. |
| 19:03.53 | c1de0x | is it a drop in replacement |
| 19:03.55 | c1de0x | ? |
| 19:04.08 | landley | http://speakerdeck.com/u/mirell/p/developing-for-non-x86-targets-using-qemu |
| 19:04.12 | landley | c1de0x: not yet. :) |
| 19:04.17 | landley | The 1.0 release should be. |
| 19:04.22 | landley | I'm using aboriginal to test it out. |
| 19:04.25 | c1de0x | i often have scenarios where i need to 'carve out space' on filesystems for things i need. |
| 19:04.33 | landley | The 1.0 release has to build both aboriginal and LFS under it. |
| 19:04.36 | landley | Otherwise it's not done yet. |
| 19:04.47 | c1de0x | (btw: i'm a professional reverser... i break shit for a living) |
| 19:05.10 | landley | c1de0x: breaking stuff is somewhere between "hobby" and "metabolic process" for me... |
| 19:05.15 | c1de0x | nice target... how close are you? |
| 19:05.29 | c1de0x | ;) |
| 19:05.30 | c1de0x | metabolic process. lol. |
| 19:05.38 | landley | I'm guestimating 6 months, but it really depends how much time I get to put in on it. |
| 19:05.58 | c1de0x | anyhow. i didn't read that deck... i just read the entire aboriginal code-base. |
| 19:05.58 | c1de0x | :D |
| 19:06.02 | landley | http://landley.net/toybox/roadmap.html |
| 19:06.20 | landley | c1de0x: the deck is about the design. :) |
| 19:06.26 | landley | _why_ I did stuff, not how. :) |
| 19:06.42 | c1de0x | oh... i think its fairly obvious (mostly) from the code... |
| 19:06.54 | c1de0x | but i'll pop that deck into my queue ;) |
| 19:12.15 | landley | The performance bits start at http://speakerdeck.com/u/mirell/p/developing-for-non-x86-targets-using-qemu?slide=209 |
| 19:12.15 | landley | It's about time for Mirell and I to redo said deck, actually... |
| 19:12.15 | c1de0x | anyhow. like i said, i often need to build tools or assemble exploits for just about every one of the targets you've covered with aboriginal... and before i had built toolchains mostly by hand. so aboriginal == godsend ;) |
| 19:12.16 | landley | Yay. Glad I could help. :) |
| 19:12.16 | landley | needs to add more targets... |
| 19:12.16 | landley | needs to ask about the status of m68k on the qemu mailing list... |
| 19:12.16 | landley | And fix sh4 so it has more than 64 megs of ram in its' board emulation... |
| 19:12.16 | landley | And dig up the microblaze toolchain patches... |
| 19:12.17 | c1de0x | yeah. i have a bunch of sh4 devices lying around. i even have one which i've been using as a 'toolchain host'... |
| 19:12.17 | c1de0x | one thing that i would love to see in aboriginal is the ability to build using eglibc or even standard glibc. |
| 19:12.17 | c1de0x | do you know segher's buildall scripts? |
| 19:12.17 | c1de0x | he uses them to build kernels and kernel-building toolchains for just about every linux arch supported. |
| 19:13.24 | c1de0x | before i met aboriginal, i was just using some extensions i hacked up to buildall to give me eglibc/uclibc toolchains for arm[bl] and mips{el,be} |
| 19:16.05 | c1de0x | btw: distcc doesn't do remote linking!? i've seen some situations where linking is a massive time-hole. |
| 19:17.06 | landley | c1de0x: if it links remotely you need distributed copies of the libraries. |
| 19:17.16 | landley | If it preprocesses remotely you need distributed copies of the headers. |
| 19:17.31 | calculus | landley: yeah, like mipsel64 (a device I am working on now) |
| 19:17.32 | c1de0x | ok i guess. |
| 19:17.36 | landley | Introduces the possibility of version skew, and the ability to leak between contexts. |
| 19:17.45 | calculus | but mipsel seems to be working for now |
| 19:17.45 | c1de0x | that makes sense... |
| 19:17.55 | landley | calculus: I have some minor mips64 support but I'm told they did a _different_ one recently... |
| 19:17.58 | c1de0x | tbh ./configure is still a bitch as well. |
| 19:18.11 | c1de0x | calculus: mipsel works for me just fine. |
| 19:18.11 | landley | c1de0x: alas, ./configure is inherently a bitch. |
| 19:18.20 | c1de0x | landley: ain't that the truth. |
| 19:18.54 | landley | The current "./configure, make, make install" infrastructure all needs to be replaced the way git displaced cvs. Unfortunately nobody's working on that... |
| 19:19.11 | c1de0x | hrm. maybe i should volunteer to help you write that distcc replacement... |
| 19:19.13 | landley | Think about make. Everybody gripes about python's significant whitespace, but make has significant _tabs_. |
| 19:19.14 | calculus | I think svn was first to displace cvs |
| 19:19.23 | landley | It mixes imperative and declarative code in the same file. |
| 19:19.28 | c1de0x | it shouldn't be that big a deal, given what it needs to do. |
| 19:19.33 | landley | Its dependency calculation doesn't take things like "menuconfig" into account. |
| 19:19.51 | c1de0x | landley: you don't need to convince me. make is borked. |
| 19:19.55 | landley | calculus: svn was "cvs done right". Git was "cvs was a bad idea". |
| 19:20.18 | c1de0x | but i don't think that that e.g. cmake are even close to being 'right'. |
| 19:20.32 | c1de0x | closer to the 'make done right' than the 'make was a bad idea' |
| 19:20.39 | c1de0x | ;) |
| 19:20.43 | landley | Agreed. |
| 19:20.52 | landley | 99% of all builds these days are "make all". |
| 19:21.05 | c1de0x | the problem is. i honestly haven't got any better ideas :( |
| 19:21.12 | landley | You should be able to go "gcc *.c" and have the compiler automatically parallelize your build for you. |
| 19:21.22 | landley | Taking advantage of SMP should not be an external tool's problem. That's step 1. |
| 19:21.53 | c1de0x | heh. i wish segher were in here. |
| 19:21.59 | landley | (Alas, to make that onework we'v egot to migrate off the gcc hairball to something that isn't maintained by an inflexible committee of zealots. Luckily there's lots of people working on that with clang, pcc, open64, and so on thanks to GPlv3.) |
| 19:22.05 | c1de0x | let me see if he's around... we can make this discussion interesting. |
| 19:22.13 | landley | has to work. |
| 19:22.18 | landley | Can we reschedule this discussion for the weekend? |
| 19:22.24 | landley | I can arrange to be here at a specific time. |
| 19:22.51 | landley | Right now I'm both still recovering from the cedar pollen attack on tuesday, and still hip-deep on work crises. |
| 19:22.58 | c1de0x | landley: uh... i don't know how easy that will be... with timezones etc. |
| 19:23.13 | landley | Pick a time? |
| 19:23.13 | c1de0x | but ping me... i'll be idling in here ;) |
| 19:23.36 | landley | I've generally been getting up at 5am central so I have time before work (because I'm a zombie afterwards). |
| 19:23.46 | c1de0x | yeah. i'm not in usa. |
| 19:23.49 | landley | Sitting in a cubicle for 8 hours makes my brain shut down even if I didn't DO anything. |
| 19:23.59 | c1de0x | hehe ;) |
| 19:24.03 | landley | 5am central is noon england, i think? |
| 19:24.25 | landley | Honestly: who thought cubicles were a good idea? |
| 19:24.37 | landley | It's claustrophobic and isolating without providing privacy or the ability to concentrate. |
| 19:24.41 | landley | It's the worst of both worlds... |
| 19:25.00 | c1de0x | yeah. i've never had the displeasure. |
| 19:25.07 | NiveusLuna | I know the claustrophobic feeling of cubicles. I work at the table instead of in one of those. |
| 19:25.16 | c1de0x | being a professional 'greyhat' means i get to be eccentric... |
| 19:25.21 | NiveusLuna | I have lots o' stuff. I don't have enough room in a cubicle. |
| 19:26.11 | c1de0x | NiveusLuna: define lots ;) |
| 19:26.29 | NiveusLuna | c1de0x: Enough that my bag of holding from thinkgeek is starting to tear. :) |
| 19:26.31 | *** join/#edev davidc__ (~davidc__@208.111.35.216) |
| 19:26.37 | c1de0x | i have two storage rooms... one of which is filled with devices i've worked on in the past... |
| 19:26.40 | landley | NiveusLuna: I'd love to be able to do that, but they pay me to be here. |
| 19:26.58 | NiveusLuna | My work is very relaxed. I work at the table that's in the room. :P |
| 19:26.59 | landley | Pay me well enough I don't complain about it, either. Not enough to retire yet, though... |
| 19:28.04 | c1de0x | heh. well. i don't wanna start griping about pay. again. |
| 19:28.31 | c1de0x | i love the work that i do, i hate the organization i do it for. |
| 19:29.41 | NiveusLuna | I love the work I do. I love the organization I do it for. Sadly I'm not being paid yet, but I should finish my product in the next two or three weeks. |
| 19:30.28 | c1de0x | lol. |
| 19:30.32 | c1de0x | NiveusLuna: are you a student/intern? |
| 19:31.30 | NiveusLuna | currently student. |
| 19:31.47 | c1de0x | just wait. one day, you too will hate the organization you work for ;) |
| 19:33.24 | NiveusLuna | I'm starting to question it already. I think our receptionist/secretary got fired for personal reasons this week rather than actual reasons. |
| 19:33.32 | NiveusLuna | she broke up with the CFO's son. >_> |
| 19:33.43 | c1de0x | :P |
| 20:22.16 | *** join/#edev factor (~factor@r74-195-184-248.msk1cmtc01.mskgok.ok.dh.suddenlink.net) |
| 20:46.58 | *** join/#edev dijenerate (~dijenerat@173.225.251.175) |
| 20:47.02 | *** join/#edev ivan` (~ivan@unaffiliated/ivan/x-000001) |
| 21:22.17 | Mirell | Huzzah, going to SF in March |
| 21:23.27 | Mirell | For http://www.erlang-factory.com/ |
| 21:33.03 | calculus | Mirell: in that case we should meet (probably in the evening or on the weekend) |
| 21:38.19 | Mirell | calculus: March 26th-30th |
| 21:49.03 | *** join/#edev segher (~segher@5ED3C8DF.cm-7-4d.dynamic.ziggo.nl) |
| 21:53.14 | calculus | evenings it is |
| 21:54.00 | calculus | I should probably be free, but I will know better as the date approaches |
| 22:25.52 | *** join/#edev rsFF (~rsFF@bl15-22-37.dsl.telepac.pt) |
| 22:32.05 | *** join/#edev florian (~fuchs@Maemo/community/contributor/florian) |
| 22:44.48 | *** join/#edev War2 (~war2@unaffiliated/war2) |
| 23:40.57 | *** join/#edev bgamari_ (~ben@physicsnat56.physics.umass.edu) |