00:00:05 -!- notostraca has quit (Read error: 54 (Connection reset by peer)).
00:00:14 <ehird> It doesn't run postinstall scripts and the tar invocation must be failing.
00:00:37 -!- pedm has joined.
00:02:04 -!- pedm has left (?).
00:02:40 -!- zzo38 has quit (Remote closed the connection).
00:02:53 <ehird> No use hacking this script up further
00:03:00 <ehird> Time to rewrite it in a way I can understand
00:03:13 -!- notostraca has joined.
00:03:50 * oerjan wonders what straca is and what is so negative about it
00:20:10 -!- notostraca_ has joined.
00:20:10 -!- notostraca has quit (Read error: 104 (Connection reset by peer)).
00:20:13 -!- notostraca_ has changed nick to notostraca.
00:32:19 <ehird> http://www.mirrorservice.org/sites/sourceware.org/pub/cygwin/setup.ini
00:32:24 <ehird> Looks pretty simple
00:33:18 <ehird> Been a while since I coded something "solid" in Python; at least I'm good at architecture.
00:36:24 -!- FireFly has quit ("Later").
00:43:47 -!- futurestack has quit ("outta here").
00:46:21 <ehird> "No one had done it before (and it is something useful), so that qualifies as "impossible" to me."
00:46:35 <ehird> I am currently doing the impossible by living one second longer than anyone else before this second.
00:53:27 -!- notostraca_ has joined.
00:53:27 -!- notostraca has quit (Read error: 104 (Connection reset by peer)).
00:53:30 -!- notostraca_ has changed nick to notostraca.
00:55:47 -!- oerjan has quit ("Good night").
00:56:16 <ehird> Anyone know how autoconf etc distinguishes function from var?
00:56:27 <ehird> int main() { (void) symbol; return 0; }
00:56:34 <ehird> is a pretty good way to check if a symbol exists
00:56:43 <ehird> but what about function/not function?
01:01:31 -!- notostraca_ has joined.
01:01:31 -!- notostraca has quit (Read error: 104 (Connection reset by peer)).
01:01:36 -!- notostraca_ has changed nick to notostraca.
01:10:58 <pikhq> ehird: See config.log; it contains the source code for the checks.
01:11:12 <ehird> I was afraid you'd say that.
01:11:20 <pikhq> Hrmm. Actually, it does that if the check fails.
01:12:46 -!- BeholdMyGlory has quit (Remote closed the connection).
01:13:08 <pikhq> It appears to be doing: int main () { return symbol(); ; return 0; } // and checking for compile errors.
01:16:22 -!- notostraca has quit (Read error: 104 (Connection reset by peer)).
01:16:24 -!- notostraca_ has joined.
01:16:43 * ehird ponders porting ghc to cygwin
01:16:52 <ehird> otherwise I can't try out darcs that integrates with cygwin
01:17:42 <ehird> Ooh, there's a Cygwin 1.7 on the way
01:17:51 <ehird> HOLY FUCK YOU CAN SEARCH PACKAGES IN SETUP.EXE
01:17:54 <ehird> All else is basically forgiven.
01:18:45 <ehird> Still stuck on bash 3, why am I surprised
01:19:59 -!- Sgeo has joined.
01:22:54 -!- notostraca has joined.
01:22:54 -!- notostraca_ has quit (Read error: 104 (Connection reset by peer)).
01:24:16 -!- notostraca_ has joined.
01:24:16 -!- notostraca has quit (Read error: 104 (Connection reset by peer)).
01:24:19 -!- notostraca_ has changed nick to notostraca.
01:29:07 -!- notostraca has quit (Read error: 54 (Connection reset by peer)).
01:29:10 -!- notostraca has joined.
01:39:13 * ehird wonders how to run ./configure without a shell :-P
01:43:17 -!- notostraca_ has joined.
01:43:17 -!- notostraca has quit (Read error: 104 (Connection reset by peer)).
01:43:20 -!- notostraca_ has changed nick to notostraca.
01:44:26 <ehird> pikhq: do you think they'll add support for windows batch? can't be worse than ksh -3.4.vendor2.release7.quirkversion.warezpatch
01:45:23 <ehird> ...no but seriously, I think I'll bootstrap my crazy endeavor to compile zsh on stock Windows with msys bash
01:46:08 <ehird> with no compatibility layers
01:46:57 <ehird> i am more insane than a flock of fish
01:47:00 <ehird> or a shoal of cows
01:47:16 <ehird> a shoal of MOTHERFUCKING cows.
01:48:41 <ehird> of course now i have to persuade configure that that isn't msys behind the curtain
01:50:04 <ehird> soon it'll hit fork
01:51:18 <ehird> actually it just wants curses right now.
01:56:02 -!- ehird_ has joined.
01:56:16 -!- ehird has quit (Nick collision from services.).
01:56:18 -!- ehird_ has changed nick to ehird.
01:56:19 -!- ehird has quit (Remote closed the connection).
01:56:34 -!- ehird has joined.
01:58:09 -!- notostraca_ has joined.
01:58:09 -!- notostraca has quit (Read error: 104 (Connection reset by peer)).
01:58:12 -!- notostraca_ has changed nick to notostraca.
02:04:00 -!- notostraca_ has joined.
02:04:00 -!- notostraca has quit (Read error: 104 (Connection reset by peer)).
02:04:03 -!- notostraca_ has changed nick to notostraca.
02:04:23 <Gracenotes> does anyone know if graphviz has been ported to a different programming language (than C)?
02:04:34 -!- coppro has joined.
02:10:57 -!- Gracenotes has quit (Read error: 145 (Connection timed out)).
02:20:42 -!- notostraca_ has joined.
02:20:42 -!- notostraca has quit (Read error: 104 (Connection reset by peer)).
02:20:47 -!- notostraca_ has changed nick to notostraca.
02:24:14 -!- coppro has quit (Remote closed the connection).
02:27:54 -!- Gracenotes has joined.
02:28:21 -!- coppro has joined.
02:32:07 -!- augur_ has joined.
02:49:56 -!- notostraca_ has joined.
02:49:56 -!- notostraca has quit (Read error: 104 (Connection reset by peer)).
02:49:59 -!- notostraca_ has changed nick to notostraca.
02:55:30 -!- encoded has joined.
02:55:42 -!- encoded has left (?).
03:11:36 -!- notostraca has quit.
03:18:49 -!- ehird has quit (Read error: 113 (No route to host)).
03:29:03 -!- adam_d_ has quit (Read error: 113 (No route to host)).
04:10:13 -!- coppro has quit (Remote closed the connection).
04:11:20 -!- coppro has joined.
04:20:39 -!- coppro has quit (Remote closed the connection).
04:21:52 -!- coppro has joined.
04:39:30 -!- Pthing has quit (Remote closed the connection).
05:29:03 -!- Asztal has quit (Remote closed the connection).
05:32:53 -!- CESSMASTER has quit (Read error: 110 (Connection timed out)).
05:43:40 <GregorR> With Op. 10 I created two betas ... with Op11, I have a "work in progress preview" :P http://codu.org/music/Op11/GRegor-op11-wipp-1.ogg
05:47:22 <pikhq> Ooooh, Gregor opus.
05:52:14 -!- Gracenotes has quit (Remote closed the connection).
05:58:11 -!- Gracenotes has joined.
06:03:19 -!- Sgeo has quit (Read error: 110 (Connection timed out)).
06:19:05 -!- coppro has quit (Remote closed the connection).
06:19:17 -!- lifthrasiir has joined.
06:43:36 -!- bsmntbombdood has changed nick to bsmntbombdood___.
07:59:59 -!- clog has quit (ended).
08:00:00 -!- clog has joined.
08:06:07 -!- kar8nga has joined.
08:34:25 -!- Gracenotes has quit (Read error: 60 (Operation timed out)).
08:52:55 -!- kar8nga has quit (Remote closed the connection).
09:19:12 -!- KingOfKarlsruhe has joined.
09:21:58 -!- KingOfKarlsruhe has quit (Client Quit).
09:22:05 -!- KingOfKarlsruhe has joined.
09:40:20 -!- adam_d_ has joined.
10:38:59 -!- oerjan has joined.
10:51:24 -!- sebbu has joined.
10:57:06 -!- Slereah_ has joined.
11:09:12 -!- sebbu2 has quit (Read error: 110 (Connection timed out)).
11:12:36 -!- FireFly has joined.
11:17:05 -!- Slereah has quit (Read error: 110 (Connection timed out)).
11:27:17 -!- BeholdMyGlory has joined.
11:42:12 <AnMaster> when was the last time you used octal numbers? And in what language?
11:50:22 <fizzie> x86 assembler, for a 'open' file-creation mode. It's oh-so-natural in that context.
11:50:55 <fizzie> nasm, to be precise; surprisingly enough it didn't interpret "0123" as octal, wanted a "0o" prefix there.
11:51:26 <oerjan> hm does that mean chmod counts as a use? although i really use a+x style
11:52:41 <fizzie> I'm not sure chmod counts as a programming language.
11:53:15 <fizzie> I don't recall last not-related-to-file-modes use of octal numbers, though.
11:54:14 * oerjan guesses there probably was some math puzzle involved, rather than actual programming (except to solve the puzzle)
11:55:41 <oerjan> "Share prices are more attuned to magic and many would visit regularly to pray." Today's mezzacotta seems related to the current crisis somehow...
11:59:10 <oerjan> `calc 18.77 parsecs in light years
11:59:13 <HackEgo> 18.77 Parsecs = 61.2209126 light years
12:04:50 <oerjan> hm interesting poll variation
12:08:17 <oerjan> "He blings an army of crones to our cause" XD
12:17:39 -!- oerjan has quit ("Later").
12:59:27 <AnMaster> anyone know unicode code point of = but with three lines?
12:59:49 <AnMaster> wasn't able to find it after several minutes of searching
13:10:31 -!- sebbu2 has joined.
13:16:27 <fizzie> It's right there in the mathematical operators block.
13:18:52 -!- sebbu3 has joined.
13:30:11 -!- sebbu has quit (Read error: 110 (Connection timed out)).
13:38:22 -!- sebbu2 has quit (Read error: 110 (Connection timed out)).
14:01:26 -!- Pthing has joined.
14:28:23 -!- sebbu3 has changed nick to sebbu.
14:40:31 -!- CESSMASTER has joined.
14:51:57 -!- CESSMASTER has quit (Read error: 104 (Connection reset by peer)).
14:54:09 -!- CESSMASTER has joined.
15:01:16 -!- ais523 has joined.
15:03:01 -!- sebbu2 has joined.
15:15:58 -!- comex has joined.
15:20:36 -!- sebbu has quit (Read error: 110 (Connection timed out)).
15:44:58 -!- KingOfKarlsruhe has quit (Remote closed the connection).
17:05:00 -!- KingOfKarlsruhe has joined.
17:10:34 -!- ais523 has quit (Remote closed the connection).
17:14:17 -!- KingOfKarlsruhe has quit (Remote closed the connection).
17:15:08 -!- ais523 has joined.
17:20:22 -!- coppro has joined.
17:36:17 -!- coppro has left (?).
17:38:18 -!- Gracenotes has joined.
17:39:01 -!- atrapado has joined.
17:50:23 -!- Asztal has joined.
17:52:00 -!- FireFly has quit (Read error: 104 (Connection reset by peer)).
17:52:46 -!- FireFly has joined.
18:11:35 <Asztal> 01:58:24 <ehird> i want a button that sodomises everyone who tries to fake native widgets repeatedly
18:12:22 <Asztal> I would like that; I could get back at the people who made the .NET MenuStrip and ToolStrip controls, which emulate that horrible "Office-style" look
18:12:37 <Slereah_> If that is so, I'll have to start faking native widgets
18:13:02 <fizzie> It's like a gadget, except more widdly.
18:13:43 <Asztal> Drawing real-looking menus on Vista+ isn't _that_ hard; I've done it.
18:35:59 -!- oerjan has joined.
18:37:01 <oerjan> <ehird> i want a button that sodomises everyone who tries to fake native widgets repeatedly
18:37:20 <oerjan> you realize that would lead some people to make a lot of them, right?
18:37:20 <ais523> what if they only try to fake native widgets once/
18:38:30 <oerjan> <Slereah_> If that is so, I'll have to start faking native widgets
18:39:07 <oerjan> STOP STEALING MY JOKES
18:40:14 -!- kar8nga has joined.
18:42:28 <oerjan> http://www.verb2verbe.com/conjugation/english-verb/snore.aspx <-- FAIL
18:43:32 <oerjan> (i was trying to look up if snore was a strong verb i didn't know about since someone on the iwc forum inflected it as such)
18:44:23 <oerjan> but, well, that thing in the link was not among the options
18:47:24 <oerjan> hm it seems to be a specific error for that word, not for other verbs ending in e
18:48:16 <fizzie> If you disobey, you will be snoreed.
19:12:11 -!- KingOfKarlsruhe has joined.
19:25:29 -!- BeholdMyGlory has quit (Remote closed the connection).
20:04:03 -!- BeholdMyGlory has joined.
20:10:01 <AnMaster> <Asztal> 01:58:24 <ehird> i want a button that sodomises everyone who tries to fake native widgets repeatedly <-- could I have one for "uses boost"?
20:11:08 <AnMaster> Asztal, what? libboost is a absolutely horrible pile of bloat and anti-design-patterns
20:11:19 <AnMaster> it is like taking the worst of C++ and making it even worse
20:11:21 <Deewiant> AnMaster: He uses boost in Stinkhorn :-P
20:11:44 <AnMaster> Deewiant, the worst part is that there is always one game or one program that you need that uses libbost
20:11:51 <AnMaster> so you can't avoid installing it
20:12:31 <Asztal> It don't actually use the libraries, though, just the headers, because boost takes hours to build
20:12:37 <AnMaster> often you end up needing a "newer than in package repo" version too, so you have to install the -dev packages for it as well (this only apply to binary distros obviously)
20:12:49 <AnMaster> Asztal, I thought C++ already did?
20:13:39 <AnMaster> Asztal, why the hell are you using the libboost headers though?
20:14:36 <Asztal> for things like shared_pointer, noncopyable, pool_allocator
20:16:19 <Ilari> Smart pointer class, make class noncopyable and better allocator if one has large number of objects for the same class.
20:16:30 <AnMaster> Asztal, oh and writing a pool allocator that allocated fixed sized chunks from large pools allocated with malloc was a 10 minute job in C. And about 40 lines of code.
20:16:59 <Deewiant> AnMaster: Yours probably isn't quite as ultra-optimized as the one in Boost, though.
20:17:32 <Asztal> oh, and lexical_cast, which is ugly and I should probably remove
20:18:14 <Asztal> basically ToString and FromString that uses string streams
20:18:28 <AnMaster> Ilari, smart pointers? Is that basically transparent ref count?
20:18:41 <Asztal> so if the data type can be written to an ostream, you can convert it to a string, and the same for istream.
20:18:48 <AnMaster> or since it is C++, I bet it isn't very transparent in practise
20:19:26 <AnMaster> Deewiant, is stinkhorn on the mycology results page?
20:20:32 <AnMaster> Asztal, stinkhorn sounds like it uses all the worst features of C++
20:21:07 <AnMaster> Asztal, also what is the point of making a class non-copyable?
20:21:18 <AnMaster> I mean, just don't do it if you don't want to
20:21:29 <Asztal> It's not that simple in C++.
20:21:40 <Ilari> AnMaster: It manages resources that don't make sense when copied?
20:22:02 <Asztal> C++ makes temporary copies of objects all over the place if you're not careful, for one
20:22:05 <AnMaster> Ilari, well ok, why copy it then? can't you keep track of what you are doing yourself?
20:24:30 <AnMaster> Deewiant, btw how could boost's pool allocator be more optimised?
20:24:46 <AnMaster> better algorithm? if so, what one?
20:24:55 <Deewiant> I don't know, and that's precisely why I'd use theirs instead of one of my own.
20:25:02 <Warrigal> I suppose I should design a language with prepositions at some point.
20:25:51 <AnMaster> Deewiant, I think Warrigal meant like "assign that to it"
20:26:31 <Deewiant> "that" and "it" are both pronouns, not prepositions.
20:26:57 <AnMaster> a language with pronouns would be nice
20:27:16 <AnMaster> and applescript already has prepositions iirc
20:28:32 <Deewiant> D has as many as "in", "for", and "with"!
20:29:18 <Deewiant> Haskell only has "in" and "of"... maybe I should re-evaluate my preferred language
20:30:06 <Warrigal> foldl1' with (>>) map (.) forkIO with hello onto names
20:30:07 <Deewiant> It's a conjunction, in the typical use.
20:30:25 <AnMaster> well, it can be a noun ("a while ago") but is it a noun in "while you are doing that, ..."
20:31:02 <Warrigal> Hmm, I don't know what part of speech it is in the subordinate clause "while you are doing that".
20:31:58 <Warrigal> You can't do that with other conjunctions. And there's a lot you can do with other conjunctions that you can't do with this.
20:32:06 <Warrigal> I guess new parts of speech are long overdue.
20:32:30 <Deewiant> "if you are doing that", "because you are doing that", etc...
20:33:07 <Deewiant> Right, presumably there are subclasses among the conjunctions.
20:33:10 <Warrigal> "I am eating some soup while a sandwich."
20:33:42 <AnMaster> Warrigal, what word would work there?
20:33:50 <AnMaster> but that is a preposition right?
20:33:58 <Deewiant> Yep: "while" is a subordinating conjunction, "or" is a coordinating conjunction << http://en.wikipedia.org/wiki/Grammatical_conjunction
20:35:42 <AnMaster> I have no idea what class that is
20:39:58 -!- augur_ has quit ("Leaving...").
20:45:39 -!- KingOfKarlsruhe has quit (Remote closed the connection).
20:53:19 <AnMaster> Deewiant, I suggest you update your mycology results page
20:53:31 <AnMaster> since it says cfunge doesn't implement NCRS
20:54:12 <Deewiant> Actually, it says cfunge 0.3.3 doesn't implement NCRS, which remains true
21:11:31 -!- zzo38 has joined.
21:16:48 -!- zzo38 has left (?).
21:20:14 -!- ehird has joined.
21:20:42 <oerjan> the hird of the sheeple
21:20:53 <ehird> 03:42:12 <AnMaster> when was the last time you used octal numbers? And in what language?
21:21:16 <AnMaster> ehird, doubtful that is a programming language
21:21:39 <ehird> It's a Unix command that takes some input and a description of what to do of it and then outputs output accordingly.
21:21:51 <ehird> Just like awk, sed, or find.
21:22:07 <ehird> cat does not take a description of what to do with the input
21:22:32 <AnMaster> that replaces control codes with ^F
21:22:57 <ehird> And GNU is Not Unix in more than implementation./
21:23:21 <ehird> AnMaster: posixcrap.
21:23:49 <ehird> AnMaster: -u is not unix, -u is a crappy hack
21:24:11 <ehird> "cat came back from Berkeley waving flags" --Rob Pike
21:24:18 <AnMaster> ehird, -u is POSIX. SuS is a superset of POSIX.
21:24:31 <ehird> Neither follow the unix philosophy.
21:24:42 <ehird> Unix is older than the bullshit pile known as POSIX.
21:24:55 <ehird> You are disqualified from EVER speaking about what is Unixy or not.
21:25:14 <AnMaster> ais523, do you know who Rob Pike is?
21:25:14 <ehird> Robe Pike, member of the original Unix team.
21:25:29 <ehird> AnMaster: ais523 isn't arguing based on historical grounds what is unixy or not
21:25:31 * ais523 tends not to care about the names of programmers
21:25:48 <AnMaster> ehird, of course unix existed before POSIX
21:26:04 <AnMaster> but these days, POSIX is very important in defining what *nix is
21:26:40 <ehird> i wish i never answered your idiotic question.
21:27:10 <AnMaster> ehird, chmod is not a programming language still
21:27:36 <ehird> you do realise there is a chmod system call, retard?
21:27:47 <ais523> AnMaster: now you say that, you make me want to try to make chmod into a programming language
21:27:50 <AnMaster> what has that got to do with it
21:27:53 <ais523> but I seriously doubt it has any power at all
21:27:59 <ais523> hmm.... maybe if you had a maze of symlinks
21:28:00 <ehird> AnMaster: THAT TAKES AN OCTAL NUMBER AS A PARAMETER?
21:28:12 <ais523> and something that iterated over a directory structure, running all the executable things there
21:28:23 <AnMaster> <ehird> AnMaster: THAT TAKES AN OCTAL NUMBER AS A PARAMETER? <-- yes
21:28:28 <ais523> AnMaster: I suspect even like that, chmod will be less powerful than Deadfish
21:28:32 <AnMaster> int chmod(const char *path, mode_t mode);
21:28:41 <AnMaster> where mode_t is in fact an integer
21:28:54 <ehird> AnMaster: and you don't see how this is relevant?
21:29:03 <AnMaster> ehird, well, it is a thin wrapper for the system call
21:29:08 <ehird> are your neurons actually firing, AnMaster
21:29:12 <ais523> I haven't been paying attention; someone summarise the arguments?
21:29:26 <AnMaster> while ls does a bit more work, with formatting and such
21:29:27 <ehird> ais523: AnMaster is dense, I'm not
21:29:36 <ehird> (this is a condensed version of the opposite summary AnMaster will give you)
21:29:49 <AnMaster> ais523, I'm still waiting for ehird to get to the point
21:30:17 <ais523> ok, the issue seems to be that in "<AnMaster> when was the last time you used octal numbers? And in what language?", ehird's answer to the first question means there's no really sensible answer to the second
21:30:17 <AnMaster> ehird, how is it more like a programming language than, cd or cat
21:30:25 <ehird> ais523: yes, there is
21:30:28 <ais523> assuming that the chmod call was from the commandline, rather than, say, from C or Perl
21:30:35 <ehird> take, for example, bash
21:30:42 <ais523> yes, bash would be another possibility
21:30:50 <ehird> that's just as valid a programming statement as chmod(file, 0o777)
21:30:53 <ehird> or whatever prefix
21:31:14 <ehird> therefore, using it in chmod(1) absolutely counts as a valid reason to use it in a programming language, especially as you can use it in bash
21:31:16 <ais523> hmm... philosophical question: would you consider typical Makefiles to be programs?
21:31:16 <AnMaster> but chmod itself is no programming language
21:31:24 <ehird> in the Make language
21:31:24 <AnMaster> ais523, which is what ehird claimed
21:31:26 <ais523> as opposed to ones designed specifically for programming
21:31:28 <ehird> which is declarative
21:31:33 <ais523> I'm inclined to think yes, too
21:31:34 <ehird> and also has really weird semantics
21:31:51 <AnMaster> ais523, isn't at least gnu make TC?
21:31:51 <ais523> is make TC, by the way?
21:32:03 <ehird> I'm pretty sure all makes are TC.
21:32:04 <ais523> my guess is no for traditional make (probably due to lack of storage), yes for GNU make
21:32:09 <ehird> Unless $(shell ) is some gnu-only thing
21:32:15 <ehird> Which it probably is
21:32:33 <ais523> I think traditional make would try to substitute the value of a variable called ( in there
21:32:43 <ais523> make's quoting rules are annoying
21:32:47 <ehird> You do $(foo) for vars in make, always.
21:32:51 <ehird> That's not a gnuism, that's a makeism.
21:33:04 <ehird> apart from things like $@
21:33:10 <ais523> anyway, time to go home
21:33:15 -!- ais523 has quit (Remote closed the connection).
21:34:55 <Ilari> WebGL... Oh great. Obnoxious websites, making previously non-security-sensitive code (majority of 3D "driver" runs inside the application) remotely scriptable, etc...
21:35:42 <ehird> Absolutely, we should stop designing technologies immediately if it turns out they can be used to be obnoxious in any way
21:36:18 <Ilari> That latter (non-security-sensitive -> remotely scriptable) might be actually worse...
21:36:36 <ehird> By driver do you mean your system's?
21:36:45 <AnMaster> Ilari, any specific security issue?
21:36:46 <ehird> Because running that in the browser is bizarre
21:37:43 <Ilari> No specific issue, but taking code that previously was not security sensitive and then making it remotely scriptable triggers my hinky detector....
21:38:10 <AnMaster> ehird, a large chuck of the opengl driver runs in libGL.so iirc. Which is inside the browser. The kernel/X bit of the driver is mostly talking to hardware and making sure more than one opengl app works side by side nicely.
21:38:26 <ehird> I'd expect libGL.so to be quite solid.
21:38:29 <AnMaster> ehird, this is all IIRC. Note that
21:38:40 <ehird> But I can't figure out how it'd be a security flaw.
21:39:04 <AnMaster> ehird, anyway keeping most stuff out of kernel is good for security (obviously)
21:39:33 <AnMaster> since X runs as root for hardware access (though that is starting to change)
21:39:54 <ehird> X is a shitpile, no point patching it up. It's insecure by design.
21:39:59 <Ilari> ehird: If code is not security sensitive, do you code very carefully to avoid buffer overflows (and even coding "very carefully" often isn't enough)?
21:40:21 <ehird> AnMaster: Relatively, compared to the rest of the system.
21:40:23 <AnMaster> ehird, at least not for nvidia libGL. I can say that much
21:40:30 <ehird> Ilari: In a driver running in a shitload of programs? Yah, I do.
21:41:37 <Ilari> ehird: Normal drivers are usually security-sensitive. libGL.so isn't.
21:41:51 <ehird> Why? Because it's used only in vidya gaems or something?
21:41:58 <Ilari> Until something (like WebGL) makes it remotely scriptable...
21:42:34 <Ilari> Because it runs inside memory space of process using it and doesn't accept external untrusted input.
21:43:14 <AnMaster> yep, most of the stuff libGL does is keeping state, consider how you do opengl programming ehird...
21:44:03 <AnMaster> (and of course, translating it in some way when you finished adding polygons, and then sending it to X server)
21:44:18 <ehird> It's as much of a security hole as anything else.
21:44:30 <AnMaster> ehird, it wasn't before. that was the point.
21:44:32 <ehird> By your argument, we shouldn't add new features to the browser: they're not coded super-carefully
21:44:38 <ehird> (@Ilari, not AnMaster)
21:44:46 <AnMaster> ehird, that is a straw man on what Ilari said.
21:44:53 <Ilari> Actually, probably sending the polygon data directly to graphics card (maybe signaling kernel to DMA the whole buffer).
21:44:59 <ehird> Learn what "strawman" is, it's the exact same argument he used
21:45:13 <AnMaster> Ilari, it probably goes through X using shm
21:45:34 <AnMaster> ehird, no it isn't exactly the argument he used
21:45:54 <ehird> Assertion match! Woohoo
21:46:05 <AnMaster> ehird, because it was about when it exposes hardware access, not new features in general
21:46:11 <Ilari> ehird: Browser (or at least many parts of it) is security-sensitive.
21:46:24 <AnMaster> adding something like canvas is quite different than canvas3d
21:46:41 <Ilari> ehird: Especially the javascript interpretter. And look at its security track record...
21:47:05 <Ilari> ehird: Non-security-sensitive code would probably be much much worse if exposed to scripting...
21:47:20 <ehird> Ilari: uh, the js interpreter is way way less solid than libgl...
21:48:43 <AnMaster> ehird, libgl hasn't been the target of exploits so far really. Because it tends to be hard to get at from a browser, unless you already used some exploit. And no server software is likely to have it loaded.
21:49:13 <AnMaster> so that "less solid" doesn't mean much
21:49:14 <ehird> Dude. Ilari is arguing that the JS interpreter is more solidly-coded than libgl.
21:49:19 <ehird> I cannot entertain such insanity.
21:49:33 <ehird> The JS interpreter in all browsers I know of gives no eye to security.
21:49:40 <AnMaster> ehird, I'm arguing that the js interpreter has been the target of a lot more exploiting attempts than libgl
21:50:11 <AnMaster> ehird, it is like saying "mac os 9 is very secure, there are only around 15 viruses for it, compared to lots for windows 98"
21:50:24 <ehird> So are you arguing for or against Ilari? I can only construe that as against
21:50:30 <Ilari> ehird: If it didn't give any eye to security, attackers would *walk* all over it.
21:51:31 <Ilari> Warrigal: Its quite insecure in practice, but attackers don't really walk all over it.
21:51:45 <ehird> Warrigal... didn't say anything.
21:51:52 <ehird> Nobody starting with a W said anything :P
21:52:01 <ehird> Well, w is next to e in qwerty.
21:52:15 <ehird> And Warrigal is the first alphabetical completion.
21:52:33 <AnMaster> I think Ilari has a point here
21:52:51 <AnMaster> that libgl hasn't been something available to exploit before
21:53:08 <AnMaster> while the js interpreter always is there
21:53:36 <AnMaster> libpng is another one that has been targeted. Same for libjpeg and so on
21:54:18 <AnMaster> but libgl has just not been something that was ever loaded into any of the apps you would target (server software, browsers, email clients, ...)
21:54:50 <Ilari> Not all exploits using JS are in javascript interpretter itself. It can also assist in exploiting other bugs... But those are still in security-sensitive parts.
21:55:11 <Ilari> libpng, libjpeg, libz, all secuirty sensitive...
21:55:32 <AnMaster> ehird, thus liubgl is less carefully reviewed by both good and bad guys for security issues than something that is easier to get at
21:55:52 <AnMaster> Ilari, yes and those have been quite well reviewed, and had a number of CVEs
21:57:11 <AnMaster> also libgl bugs are common. but so far, the worst that led to was misrendering, or in worst case, segfault
21:57:16 -!- kar8nga has quit (Remote closed the connection).
21:57:29 <AnMaster> ehird, I would not call libgl solid
21:57:39 <ehird> more solid than a JS impl.
21:57:41 <Ilari> Well, if some piece of code is written in language where buffer overflows are possible (without exploiting the runtime), don't use extraordinary measures to be buffer-overflow-free, function in security-sensitive function and don't have CVEs, then probably they haven't been audited throughly enough.
21:57:42 <ehird> Anyway, the v1 iPhone OS was jailbroken with a hole in... libjpeg? Or libtiff or something.
21:58:21 <ehird> You went to jailbreakme.com and it showed an image; it had the invalid image/404 icon on it, but your phone froze, a progress bar came up, and jailbreaked it.
21:58:27 <AnMaster> Ilari, and there is more than one libGL. My laptop and desktop use different ones
21:58:38 <AnMaster> generic open source one, and nvidia one
21:59:24 <AnMaster> ehird, what is the point you are trying to make here? it seems you are arguing for Ilari now.
21:59:35 <ehird> I was just commenting.
21:59:58 <AnMaster> I run browser as a different user account
22:00:10 <Ilari> In this case, that diversity probably doesn't protect that much, because attack doesn't need to commit on one... And there are probably vendor strings available anyhow...
22:00:28 <ehird> I find it possible to be surprised. You probably run everything as a different user account, in case someone exploits your IRC client.
22:00:35 <AnMaster> Ilari, my point was that nvidia one is a binary blob
22:00:53 <ehird> One wonders how the fuck you expect people to exploit you browser more than any other piece of software if you disable JS.
22:01:10 <AnMaster> ehird, you just provided an example
22:01:19 <ehird> "More than any other piece of software"
22:01:26 <ehird> No part of AnMaster's system apart from his browser displays images.
22:01:30 <AnMaster> ehird, and yes irc is a different acount.
22:01:36 <Ilari> BTW: The standard-issue browser doesn't *support* JS. It just doesn't have it disabled, it just plain does not support it.
22:01:51 <Ilari> *standard-issue browser I use
22:01:56 <ehird> AnMaster: cool, it's not even possible to satirize you.
22:02:10 <Ilari> Links2, both in text and graphics mode.
22:03:05 <ehird> nc(1) has no security flaws, at least as a client.
22:03:15 <ehird> Nor does sed if you want prettier pages.
22:03:42 <AnMaster> ehird, how do you know nc has no buffer overflow?
22:03:53 -!- atrapado has quit ("Ex-Chat").
22:04:14 <AnMaster> oh and, the sed script must be reviewed for faulty code
22:04:25 <ehird> Because nc is some 6,000 lines of sharp code.
22:04:29 <AnMaster> not buffer overflow, of course, but otherwise bad code
22:04:34 <Ilari> AnMaster: AFAIK, there is.
22:04:34 <ehird> A lot of that is server-based.
22:05:33 <Ilari> Or at least, there have been security bug reports against Links2.
22:05:36 <ehird> 12:10:01 <AnMaster> <Asztal> 01:58:24 <ehird> i want a button that sodomises everyone who tries to fake native widgets repeatedly <-- could I have one for "uses boost"?
22:05:38 <ehird> Boost isn't that bad.
22:06:00 <ehird> Ilari: I think he thought you meant there was a buffer flow in nc
22:06:13 <ehird> 22:03<AnMaster>ehird, how do you know nc has no buffer overflow?
22:06:15 <ehird> 22:04<Ilari>AnMaster: AFAIK, there is.
22:06:17 <ehird> 22:04<AnMaster>Ilari, heh
22:06:21 <AnMaster> <ehird> Ilari: I think he thought you meant there was a buffer flow in nc <-- no
22:06:43 <ehird> Apparently you're telepathic
22:07:16 <AnMaster> ehird, it followed from the context of highlights
22:07:33 <ehird> Yep, nobody ever replies to a message not directed at them
22:07:41 <ehird> No matter how old the highlight it is it must be that
22:07:59 <AnMaster> ehird, now you are just silly, it was crystal clear from the context what was intended
22:08:20 <ehird> No, it wasn't, because it was non-obvious to me, for instance.
22:08:30 <Deewiant> If a message makes sense as a response to the latest highlight, assuming it was intended for that is reasonable.
22:08:38 <ehird> I don't find that true in practice.
22:08:39 <AnMaster> ehird, that says more about you, than about the context.
22:08:40 <ehird> 12:16:30 <AnMaster> Asztal, oh and writing a pool allocator that allocated fixed sized chunks from large pools allocated with malloc was a 10 minute job in C. And about 40 lines of code.
22:08:48 <ehird> yes, why not waste 10 minutes to produce non-optimal code
22:09:00 <ehird> using an optimised ready-made version? inconceivable
22:09:16 <AnMaster> ehird, well note C vs. C++ for example.
22:09:21 <Ilari> Reminds me of one program I wrote. Immediately after receive buffer there was PROT_NONE page. So any overflow of receive buffer is instant SIGSEGV (and the program had no SIGSEGV handler)...
22:09:26 <ehird> 12:20:32 <AnMaster> Asztal, stinkhorn sounds like it uses all the worst features of C++
22:09:33 <ehird> you seem to be defining that as "everything apart from classes"
22:10:12 * AnMaster puts ehird on ignore for now, until the nonsense log reading comments finished
22:10:23 <AnMaster> if you read on, you will see that Asztal agreed for example
22:10:29 <ehird> Good idea. You might read something you disagree with.
22:10:53 <ehird> Oh, I see! Asztal made a statement supporting "stinkhorn using the worst features of C++", therefore my comment about the surrounding context in which you seemed to dislike everything that wasn't classes must therefore be false.
22:11:17 <ehird> 12:22:05 <AnMaster> Ilari, well ok, why copy it then? can't you keep track of what you are doing yourself?
22:11:25 <ehird> ...and that's why AnMaster defines every type in his C program as int
22:12:34 -!- coppro has joined.
22:15:14 <Ilari> BTW, if you want to see horrible C++ code, look at Firefox code. Not using those "horrible" C++ features results quite horrible code...
22:15:55 <Ilari> I think Firefox does it because bad C++ compilers don't support pretty much anything on top of C besides classes...
22:15:56 <ehird> Breaking your mind and shit yeah
22:17:07 -!- augur has joined.
22:17:34 <Ilari> If you want to really test C++ compiler, test wheither it survives some tricky partial template specialization. ICEs and segfaults ahead if the compiler isn't really stable...
22:17:56 <pikhq> Firefox's source code is pretty much entirely awful.
22:18:17 <Ilari> AnMaster: IIRC, at least some time ago (probably still), it didn't even use exceptions(!).
22:18:19 <pikhq> I shudder to think that that's the result of a *rewrite* of it...
22:19:03 <AnMaster> pikhq, eh? when was it rewritten?
22:19:15 <pikhq> AnMaster: It was originally Netscape 4.
22:19:25 <Asztal> It still doesn't use them, AFAIK, but they're working on automated rewriting for that
22:19:31 <pikhq> The Mozilla Project rewrote the entire thing.
22:19:46 <AnMaster> pikhq, ah that far back. yeah that was probably horrible.
22:20:25 <AnMaster> pikhq, rewrote each part one at a time? or started from scratch entirely?
22:20:53 <pikhq> Started from scratch.
22:21:00 <ehird> Netscape 3's code was altogether better, from what I hear.
22:21:03 <AnMaster> then what was the point of having the netscape code there at all?
22:21:08 <ehird> Netscape 4 is massively a result of the second system syndrome.
22:22:56 <pikhq> Pretty much the only bit of code that remained was Netscape's "NGLayout" code. NGLayout was Netscape's attempt to rewrite the layout code.
22:22:57 <Ilari> And C++ is not just C with classes. The proper programming style is quite different.
22:23:01 <pikhq> It went on to become XUL.
22:24:42 <Ilari> And why does Firefox seem to be the most bloated and resource-hungry web browser? Konqi even with all of its KDE stuff seems lighter...
22:25:03 <pikhq> Because it's poorly written.
22:26:05 <ehird> Chrome is double-plus good.
22:26:07 <AnMaster> Ilari, I use konqueror usually, unless some site doesn't work well in it
22:26:19 <ehird> ...but uses quite a bit of memory due to the process-per-tab thing.
22:26:35 <ehird> Midori or something if you want more straight WebKit without that (and also without the fast JS), I guess.
22:26:41 <Ilari> How much dirty memory does Chrome use?
22:27:14 <ehird> Ilari: Define "dirty"; I don't think Windows has a notion of it
22:28:01 <Ilari> Pretty much all OSes with virtual memory have notion of clean and dirty memory...
22:28:36 <Ilari> Dity memory => Pages that have been written and thus can't be discarded from memory (but must be swapped out).
22:28:47 <ehird> But, real memory: private 229,732kiB + shared 5,431kiB = 235,163kiB; nine tabs (one is a popup from another tab and interacts with JS stuff so is in the same process) and one process for the browser.
22:28:59 <ehird> Those pages are all quite heavy.
22:29:12 <ehird> Virtual memory: private 335,984kiB, mapped 100,204kiB.
22:29:24 <ehird> All direct quotes from about:memory.
22:29:27 <ehird> Well, with formatting.
22:29:43 <ehird> I can't measure it in Task Manager, as they're all separate processes.
22:30:20 <ehird> The values range for each process from around 18MB to 53MB. I guess that the ~3.5MB one is the browser process.
22:30:33 <pikhq> Konqueror's not a stellar web browser, but it manages to be much less bloated than Firefox simply by merit of *making design decisions*.
22:30:33 <pikhq> Firefox is a good example of feature creep.
22:31:17 <ehird> Chrome's nice and simple; the only interface apart from a tabbar and back/forwards, refresh, bookmark, the bar (both address and search) and a go button is two icon menus: a document and a wrench.
22:31:33 <ehird> The document is zoom, copy/paste, encoding, developer tools, etc: page stuff.
22:31:47 <ehird> The wrench is new tab, full screen, about, exit, options, etc: browser stuff.
22:32:08 <ehird> It's very well designed.
22:33:16 <Ilari> Its possible to make a process that would have private memory >1GiB (mostly residen't or not), but only few pages (~10KiB) of dirty memory.
22:33:25 <ehird> Still, it's not like the memory usage is correlated with bad performance; having the fastest rendering and JS engine and using native Windows widgets (albeit restyled) it's very fast.
22:33:35 <ehird> Ilari: Probably all the tabs I haven't used are paged out.
22:33:42 <ehird> *add recently after used
22:33:55 <Ilari> Dirty memory is dirty memory even after its paged out.
22:34:06 <ehird> I'm not a memory system guy, so I wouldn't know./
22:36:26 <Ilari> If you have N copies of some process, with everything resident, the memory usage would be union of all clean memory plus sum of all dirty memory.
22:36:46 <ehird> Well, chrome does note the shared memory total.
22:37:00 <ehird> Also, those totals include about:memory.
22:37:20 <ehird> Anyway, 5,250kiB of the 249,710kiB non-virtual memory is shared, according to it.
22:38:24 <Ilari> At least on Linux, shared libraries are mapped as private mappings. But all clean (non-dirty) pages are shareable.
22:39:07 <ehird> Windows probably just uses shaman magic.
22:39:52 <Ilari> Taking qbittorrent process I have running here. Its VMSIZE is ~163M, shared is ~4M. But only ~30M of its memory is dirty.
22:39:57 -!- adam_d_ has quit ("Leaving").
22:44:09 <Ilari> The private/shared is not accurate measurement of how much memory two processes share and how much they don't.
22:44:33 <ehird> True, but I can't give you better without either bloating my system with software to measure it or hacking up something myself, sorry.
22:45:04 <Ilari> Yeah, windows doesn't have equivalent of Linux /proc/X/smaps
22:47:28 <Ilari> (list of VMAs process has plus memory usage information for each).
22:49:41 <ehird> Say, does apt have pluggable backends or do you need only one server format/type/etc, I wonder.
22:50:00 <ehird> Instead of hacking up my own tool I could write an apt module backed by the cygwin package lists...
22:50:29 <Ilari> At least, apt can be pointed to servers on net and to CD-type media...
22:50:40 <ehird> But the backing format it looks at is the same there.
22:50:51 <coppro> I believe it's designed for dpkg format, yes
22:50:54 <ehird> I could write a script to convert it on a server, but "meh".
22:51:00 <ehird> coppro: That's the package format, not the package list.
22:51:18 <ehird> .debs are just tars, and so are decompressed cygwin packages, so I could easily hack that up client-side.
22:51:26 <ehird> But the package list is quite different, I think.
22:51:32 <coppro> but I mean the package list is a specific format
22:51:45 <coppro> it's the format of the Debian control file
22:51:56 <ehird> But apt is quite modular, isn't it?
22:52:51 <ehird> Obviously it's impractical for me to maintain a mirror of all Cygwin packages in .deb format, but I guess I could handle converting the cygwin.ini.
22:53:21 <ehird> But apt passes its packages off to dpkg, whereas it handles the package lists itself, so I expect exactly the wrong things are modular. Grr.
22:54:22 <coppro> I've hacked together my own tool for this before, it's not hard
22:54:46 <ehird> Converting every single Cygwin package to .deb?
22:55:00 <ehird> I'm not maintaining and serving thousands of large files on a server .
22:55:20 <coppro> for dependency analysis in a client-side package manager
22:55:42 <ehird> Oh, you mean an apt replacement.
22:55:53 <ehird> Yes, I can write my own tool; that's what I was doing. It will not be as flexible, powerful or comfortable as apt.
22:56:10 <coppro> I wouldn't classify it as a replacement, but yes
22:56:30 <coppro> you could just steal the apt source
22:56:33 <ehird> Cygwin IS basically a Linux system, and Linux systems deserve apt.
22:56:48 <ehird> coppro: Of course I'll have to patch apt. I just hope that it's modular, so I can only patch the relevant bits.
22:57:12 <ehird> That'll be fun, actually. Install dpkg with Cygwin, dpkg -i apt-cygwin.deb, vi /etc/sources/apt.list.
22:57:22 <ehird> (apt-get update, etc.)
23:04:26 <ehird> Does anyone know how to tell Sumatra PDF to reset the settings for this document to my global default?
23:15:42 <ehird> I wonder, if there are things like bb4win, are there tiling window managers for windows?
23:16:37 <AnMaster> "An assault weapon is a political [1][citation needed] term" <-- err ok. Whatever.
23:16:56 <ehird> You're not Finnish, stop that.
23:17:14 <ehird> AnMaster: Also, you're still here.
23:20:07 <ehird> Incidentally, Windows has hover-to-focus built-in.
23:24:28 -!- Slereah_ has quit (Read error: 54 (Connection reset by peer)).
23:24:30 * ehird writes a little build system for a fake program using sh to configure a config.mk file...
23:24:34 <ehird> To see if it works.
23:25:27 -!- Slereah has joined.
23:27:44 -!- fizzie has quit (Read error: 60 (Operation timed out)).
23:30:34 <ehird> http://www.windowsizer.com/ ;; would be nicer if it was one of the linux ones ported. Still downloading to try out.
23:33:33 -!- fizzie has joined.
23:33:46 <ehird> Eh, that thing doesn't auatomatically tile.
23:37:31 <ehird> Meh, I'm going to maintain my own set of cygwin .debs
23:37:35 <Asztal> how is it any better than 'Show windows side by side' on the task bar then? :|
23:37:50 <ehird> Asztal: It can do more arrangements and has preferred arrangements and stuff.
23:38:06 <ehird> But it certainly doesn't automatically handle resizes... hmm... maybe it doesn't work on >XP
23:38:44 <ehird> reading the site, almost certainly
23:38:52 <Asztal> Given that it says "Re-sizing one window automatically resizes adjacent windows", sounds broken
23:40:07 <ehird> hmm... if a deb is just an .ar with some special stuff and a cygwin package is just a .tar.bz2 with some special stuff, it can't be too hard to convert them
23:40:09 * ehird installs cygwin 1.7
23:40:21 <ehird> ...and uses the Experimental package versions
23:40:24 <ehird> they're still ancient, but eh
23:41:17 <ehird> I'm not even sure anything is tagged experimental
23:50:51 <ehird> Exp tags everything as "Skip". Fuck that, then.
23:51:13 <ehird> So Cygwin 1.7 "current" is the most up-to-date Cygwin you can get.
23:51:27 <ehird> Now featuring bash 3 plus 0.0.0.1 extra from the past release.
23:52:04 <pikhq> Apt is rather easy to patch, I *believe*.
23:52:15 -!- coppro has quit (Read error: 60 (Operation timed out)).
23:52:27 <pikhq> And that's definitely a good idea; Cygwin's setup.exe sucks.
23:52:42 <ehird> All of Cygwin sucks, it's a shoddy system.
23:53:05 -!- Gracenotes has quit (Remote closed the connection).
23:53:17 <ehird> But the only alternative, MSYS, is just a really old fork of Cygwin and sucks more (to compile executables for it you need to use a patched gcc 2; nobody bothers to update it because MSYS is supposed to be just a tool with which to run MinGW)
23:54:20 <Asztal> GNU bash, version 2.04.0(1)-release (i686-pc-msys)
23:54:47 -!- Gracenotes has joined.
23:56:05 -!- coppro has joined.
23:56:31 <ehird> Especially annoying is that no terminal program for Windows has the balls to try and emulate cmd.exe properly, so for calling out to Windows command-line programs you have to use cmd.exe.
23:56:51 <ehird> Oh, and MSYS has one huge advantage over Cygwin: it automatically translates Unix pathnames if you call a Windows tool.
23:57:06 <ehird> None of this $(cygpath -xyzzy ...) bullshit.
23:57:52 <coppro> isn't msys deprecated or something?
23:58:19 <ehird> 23:53<ehird>But the only alternative, MSYS, is just a really old fork of Cygwin and sucks more (to compile executables for it you need to use a patched gcc 2; nobody bothers to update it because MSYS is supposed to be just a tool with which to run MinGW)
23:58:21 <ehird> 23:54<Asztal>GNU bash, version 2.04.0(1)-release (i686-pc-msys)
23:58:23 <ehird> 23:54<ehird>Fuck yeah.
23:58:45 <ehird> But it's still supported and a prominent feature of the site, yes.
23:59:15 -!- BeholdMyGlory has quit (Read error: 54 (Connection reset by peer)).
23:59:44 <ehird> Really, I just want bash, coreutils like stuff, vi, rxvt and a few other things.