00:09:06 -!- zzo38 has joined. 00:09:36 HOW TO ANNOY A PHYSICIST 00:09:39 Slereah_: what are quarks made of 00:09:56 Yes I'm sure abs(i)=-1 00:10:16 Some physicists might say quarks are made of strings, I guess? 00:10:20 zzo38: depends on your definition of abs 00:10:31 zzo38: what are strings made of :D 00:10:49 And I have done pseudo-math stuff before 00:11:02 Strings are made of spacetime 00:11:05 ehird : cat guts 00:11:29 Slereah_: so.. it's a fractal universe? 00:11:37 Cat guts are made of [...] quarks which are made of cat guts. 00:11:46 And absolute value is defined as the distance from zero (at least that is how I define it and I have seen that definition but only once, and not in a book that discusses complex numbers) 00:11:59 I like my universes recursive 00:12:24 Sometimes I think of the universes recursive also, but not always 00:13:12 I will describe the INTERCAL program I wrote in case you can't read it: 00:13:42 how is the distance of i from zero -1? 00:13:45 The DO |1 <- #0 DO |2 <- #100$#0 means to make |1 refer to a new zero qubit and |2 to refer to a new one qubit 00:13:58 The distance of i from zero is positive 1. 00:14:14 Therefore abs(i)=+1 00:14:17 zzo38: it applies to complex numbers as well, just use pythagoras: abs(a+b*i) = sqrt(a^2+b^2) 00:14:18 Not -1 00:14:29 zzo38: Yes I'm sure abs(i)=-1 <<< okay this just confused me a bit 00:14:41 I already knew it applied to complex numbers as well and have used pythagoras. 00:14:57 oklofok: That was an error I meant abs(i)=+1 00:15:18 The lines DO TRANSFORM |1 DO TRANSFORM |2 applies the Hadamard transform to each register 00:15:34 (13000) is the line label for the quantum black box function 00:15:42 And then it Hadamard transform again 00:15:55 oh, you made a truly quantum INTERCAL, rather than the one in CLC-INTERCAL that's just disguised multithreading? 00:16:07 oklofok: +1 = -1 in any field of characteristic 2 *ducks obscurely* 00:16:08 And then it sets .1 to zero, and .1 to one if |1 measures to one 00:16:16 AFK 00:16:26 oerjan: like the field of one-bit numbers? 00:16:39 And then the DO READ OUT .1 since it is not a quantum command, will force a measurement and output zero or one (in roman numerals, of course). 00:16:43 ais523: that would be the prime field of c. 2, yes 00:17:09 -1i = 1i right? I'm rusty on corner cases. 00:17:09 And yes, I made (but not implemented yet!) a truly quantum INTERCAL, rather than the fake one in CLC-INTERCAL. 00:17:34 that's worrying enough 00:17:39 ehird: erm not in the complex numbers 00:17:40 I don't think I've used | in C-INTERCAL 00:17:48 if I can think of a way to implement it, I'll have to try 00:17:49 but not right now 00:17:51 oerjan: er right 00:18:10 In CLCLC-INTERCAL the | operator has a different meaning than it did in C-INTERCAL (if it had a meaning at all in C-INTERCAL) 00:18:39 it doesn't have a meaning in C-INTERCAL 00:18:46 in CLC-INTERCAL, it's one of the TriINTERCAL operators, I think 00:18:48 can't remember which one 00:19:17 ais523: (TriINTERCAL) Yes that's what I think too. But CLCLC-INTERCAL doesn't have TriINTERCAL so you don't have to worry about that 00:19:35 stop breaking sidewards compatibility! 00:20:07 If you want compatibility, just run it in compatibility mode!!!! 00:20:47 And what's "sidewards compatibility" anyways? I have only heard of backward and forward but not sideways 00:21:33 compatibility between unrelated impls of the same language 00:21:52 the important thing here should be snidewards compatibility 00:21:57 oerjan: yes i know that much field theory 00:22:14 you must support all the inside jokes 00:22:17 and much more by next friday 00:22:20 Well, there's compatibility if you run it in compatibility mode. Even FreeBASIC also has a compatibility mode and non-compatibility mode (and I use both modes of FreeBASIC). 00:22:21 yes 00:22:23 :DD 00:22:33 does freebasic let you poke to graphics memory 00:22:35 oerjan: that's a great pun, a really great pun 00:22:46 SCREEN 13 \n POKE x+(y*320), colour 00:22:46 :D 00:22:50 err 00:22:52 can I use that in C-INTERCAL advertising? 00:22:52 add a def seg in there 00:23:03 it's one of the best puns I've seen yet, and it's on-topic 00:23:10 ais523: someone should buy a google adwords for intercal 00:23:12 FreeBASIC doesn't support DEF SEG, you have to use absolute addresses without segments 00:23:14 for "programming" or something 00:23:16 ais523: i would be honored, i had no idea it was that good 00:23:22 oerjan: it wasn't 00:23:34 * oerjan swats ehird -----### 00:23:39 ERROR HANDLER PRINTED SNIDE REMARK 00:23:44 STOP PULLING ME DOWN TO EARTH 00:24:03 -!- neldoreth has quit ("Lost terminal"). 00:24:29 Yes maybe CLCLC-INTERCAL should have a "snidewards compatibility" mode as well just to be weird, whatever that means (does it have something to do with printing snide remarks or something like that?) 00:24:44 * oerjan suddenly envisions hackwards compatibility 00:24:47 zzo38: read up on INTERCAL error messages sometime 00:24:56 must ... complete ... the set 00:25:02 oerjan: compatibility with unofficially modified versions of itself? 00:25:08 maybe. 00:25:58 hmm 00:26:00 Ya maybe even compatibility with unofficially modified version of itself (but only if you enable that option) 00:26:23 oerjan: I can't think of a way to complete the set 00:26:25 but if anyone can, it's you 00:26:31 zzo38: you should implement CLCLC-INTERCAL by making a library specifically for options 00:26:34 and writing all the rest as options 00:27:01 like, --check-politeness, --constant-fold 00:27:04 for- is such a boring prefix to rhyme 00:27:09 and the options depend on other options for stages as the compiler 00:27:47 ehird: CLC-INTERCAL is implemented similarly to that anyway 00:28:00 ais523: not really 00:28:03 it's not entirely based on options 00:28:06 * ais523 finds it amusing that whenever anyone tries to think of a /really bad/ compiler design, CLC-INTERCAL already has it covered 00:28:14 ehird: the options just change which compiler files are linked in 00:28:17 so arguably, yes 00:28:19 ais523: with this, for instance, 00:28:22 you could do 00:28:27 -!- neldoreth has joined. 00:28:31 there are default option sets, but they're in an rc file 00:28:33 That's good way, maybe have options depend on other options for stages as the compiler. 00:28:44 icick --compile --no-compile-assignments 00:28:51 which would turn on compile and its dependencies but turn off --compile-assignments 00:28:57 so it'd work exactly as normal but you couldn't assign to anything 00:29:21 ehird: CLC-INTERCAL could trivially be modified to do that 00:29:22 Except that in CLCLC-INTERCAL it uses the old type of options that 1972 INTERCAL used (see the wiki page for details, and checking politeness is one of them). 00:29:23 option dependencies would be function calls 00:29:24 pretty much 00:29:27 some commands, like NEXT are in their own preloads already 00:29:30 = options 00:29:32 ais523: it's a matter of implementation 00:29:42 it's just a matter of splitting apart the files to be finer-graned 00:29:47 *grained 00:29:56 in mine, the whole compiler is one huge block simply using a massively bloated option parser 00:29:59 Yes both CLC-INTERCAL and CLCLC-INTERCAL could make assignment stop working, either DO ABSTAIN FROM CALCULATING or modify the syntax so that the assignment command does nothing 00:30:06 and instead of calling functions it depends on arguments in just the right way 00:31:25 And you modify syntax at *RUN-TIME ONLY*, not compile-time. In CLC-INTERCAL you could do both but in CLCLC-INTERCAL you can do so only at run-time. However, if you want to do all the changes at once you could CREATE a syntax that does all the other changes at once and then add the command that was just created to the end of that file. The next file in the stack will then be executed using the new compiler 00:31:50 the whole point with CLC-INTERCAL is you can't tell compile time from runtime 00:31:52 hm maybe forkwards compatibility 00:32:11 oerjan: not bad, although that's much the same as sidewards probably 00:33:23 maybe furwards or firwards for the environmentalists 00:33:27 INTERCAL compilation idea: Just Late Compilation 00:33:40 In CLC-INTERCAL I think you can decide whether to change syntax at compile-time or run-time by using crawling-registers, but CLCLC-INTERCAL has no crawling-registers 00:33:48 ehird: :D 00:33:49 It interprets the program and compiles it simultaneously. Once it's compiled a part, it rewinds the program up to that point and runs the faster compiled version. 00:33:54 ehird: awesome 00:33:55 ehird: Just-Too-Late is Claudio's name for CLC-INTERCAL's compilation method 00:34:00 ais523: haha, really? 00:34:05 :D 00:34:14 it works by running the program until it gets an error, then compiling the bit that errored and trying again 00:34:29 the error is typically a syntax error due to the code in the section not having been parsed 00:37:08 But in CLCLC-INTERCAL how the compiler should probably work, is each time a file is load or the command CREATE or DESTROY or IMPORT is used, it has to recompile the current file. 00:37:23 yes, that's boring 00:37:30 neither C-INTERCAL nor CLC-INTERCAL work like that 00:37:37 for instance, C-INTERCAL does all the compilation at compile time 00:37:45 things that don't have a meaning yet are speculatively compiled 00:37:50 so that they can run if they're given a meaning 00:37:59 DESTROY should be CREMATE, just for the confusion 00:38:15 well, C-INTERCAL has CREATE but not DESTROY 00:38:32 ao 00:38:33 so 00:38:35 I'm stark raving mad. 00:38:47 this much istrue 00:39:12 ehird: i would quote lewis carroll but people started complaining about it 00:39:14 But isn't CREATE in C-INTERCAL different than the CLC-INTERCAL and CLCLC-INTERCAL? 00:39:30 yep, it's limited to things that the compiler has a chance of guessing at with no context 00:39:35 and if anyone wants to know why I'm stark raving mad, ask in /msg because it's too mad for this place (no, really.) 00:39:36 so there are various grammar restrictions on it 00:39:49 ehird: maybe later, I have to go home soon 00:39:58 also, that's probably the second time you've mentioned it here 00:40:09 but if it's madder than creating your own email client, probably I don't want to know 00:40:15 -!- BeholdMyGlory has quit (Remote closed the connection). 00:40:29 I can't ask in /msg there is no such nick/channel 00:40:31 wait, that's mad? 00:40:38 zzo38: heh 00:41:06 And I have created my own email client before but now I don't use email anymore 00:41:12 ehird: are you sure it's too mad for here? 00:41:16 is that even physically possible? 00:41:16 ais523: maybe 00:41:22 zzo38: you have no email account? 00:41:23 -!- ais523_ has quit ("http://www.mibbit.com ajax IRC Client"). 00:41:26 *psychically possible 00:41:36 also, that took ais523_ a while, given the computer was forcibly shut down 00:41:40 about 4 minutes ago 00:46:01 I changed the CLCLC-INTERCAL wiki page so that it says that CREMATE is the new name for DESTROY (if you don't like it, you can create the new syntax for DESTROY so that DESTROY works as well) 00:47:42 ehird: Yes I have no email. But sometimes it requires email to register for something so I just wrote my own SMTP server and run it only when I am expecting one of those messages, and then cancel the SMTP server afterward 00:47:51 that's awesome :D 00:47:56 I couldn't live without email though 00:48:01 e.g. mailing lists 00:48:08 that is really quite an impressive way to get a single-use email address 00:48:25 zzo38: what should people use instead to contact you? Push-gopher? 00:49:36 There is currently no real way. If you are a member of any message boards I go on, you can send a private message there. You can also post a comment on ChronoJournal. And if you know my house address or telephone number (neither of which I will tell you) then you can use that 00:50:14 zzo38: you should invent a way that somehow lets people send messages to you via the finger protocol 00:50:25 that would be ridiculous and nobody could use it, which is what you want in a message service 00:51:12 someone should invent push-gopher now 00:51:14 O that's easy to write a way to send messages by finger protocol, just put the message as the query. And it can easily be sent too, with netcat or any finger client. But I don't currently plan on doing this 00:51:20 along the same lines as PTTH 00:51:28 What's push-gopher supposed to do anyways? 00:51:45 And what's PTTH 00:51:50 like gopher, but you send someone else to your gopher page, rather than visiting theirs 00:52:02 and PTTH is a new thing coming out of the IETF, it's like HTTP but in reverse 00:52:11 PTTH was designed for second life 00:52:13 I believe. 00:52:29 don't you mean PFFFTH 00:52:44 zzo38: O that's easy to write a way to send messages by finger protocol, just put the message as the query. <-- that's boring though, it should be like, accessing two finger sites dit and dot which are interpreted as morse code 00:52:45 O. I understand now. I could implement that if I wanted to, I guess. 00:53:42 O, you want to use morse code? 00:53:50 I guess you can use morse code if you want to. 00:54:22 Send a voice-mail to someone with morse-code so that the computer can print out your voice-mail. Someone once asked "How do I print my voicemail?" 00:54:51 zzo38: i was thinking like 00:54:59 finger dit.zzo38computer.cjb.net 00:54:59 finger dit.zzo38computer.cjb.net 00:55:00 finger dit.zzo38computer.cjb.net 00:55:03 finger dot.zzo38computer.cjb.net 00:55:06 finger pause.zzo38computer.cjb.net 00:55:08 then you'd do 00:55:10 finger end.zzo38computer.cjb.net 00:55:10 to stop 00:55:12 -!- ais523 has quit (Remote closed the connection). 00:56:03 Interesting for hypothetical esoteric programming stuff, but not for real use 00:56:31 zzo38: you could make an smtp server that takes a message and converts it to morse code then does that 00:57:10 Yes, that's possible, but I have no plans to do that 00:57:36 And next time someone asks you how to print their voice-mail, ask them if they receive their voice-mail in morse-code 00:58:21 :D 01:04:18 People can already send message to me by connecting to my HTTP site and typing a message in the URL, it will be logged, however I hardly ever read the server logs and have no guarantee that I will ever read your message. 01:04:44 :D 01:04:57 zzo38: i left you an http message 01:05:41 Thanks I found it. But that's only because you told me about it 01:06:14 maybe if we do it as a subtle DOS attack... 01:06:35 Don't do the DOS attack... 01:07:02 But sending me a message by HTTP is a good way to do it in case someone is communicating with me on a live public service where private messages are not possible. 01:07:13 But I don't know any other circumstance in which it would be useful. 01:08:12 -!- olsner has quit ("Leaving"). 01:11:07 -!- zzo38 has quit. 01:38:55 -!- oerjan has quit ("Bedwards"). 01:42:35 -!- bsmntbombdood has quit (Read error: 110 (Connection timed out)). 01:45:40 -!- FireFly has quit ("Later"). 01:57:01 -!- Nevar has quit (Read error: 104 (Connection reset by peer)). 01:58:29 -!- bsmntbombdood_ has joined. 02:34:57 "BigDecimal(): Not a number: 1" 02:34:59 Well, er... 03:00:15 Irony in command line arguments: --ignore-case and --IGNORE-CASE do different things. 03:00:43 kerlo: :D 03:03:41 Is that a joke, or serious? What do they do? 03:04:15 Those are arguments that less takes 03:04:40 One ignores case when the search or whatever contains only lowercase letters; the other ignores case in all cases. 03:04:51 (Lol, "in all cases".) 03:05:28 -!- bsmntbombdood_ has changed nick to bsmntbombdood. 04:16:41 -!- neldoreth has quit (Remote closed the connection). 04:16:52 -!- neldoreth has joined. 04:25:59 -!- cherez has joined. 04:26:09 -!- cherez has left (?). 06:02:44 -!- GreaseMonkey has joined. 06:35:29 -!- Sgeo has quit ("Leaving"). 07:12:01 -!- bsmntbombdood has quit (Read error: 110 (Connection timed out)). 07:12:49 -!- bsmntbombdood has joined. 07:14:08 -!- psygnisfive has quit (Read error: 110 (Connection timed out)). 07:29:11 oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo 07:35:29 mmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmm 07:59:59 -!- clog has quit (ended). 08:00:00 -!- clog has joined. 08:52:34 -!- Slereah_ has quit (Read error: 60 (Operation timed out)). 08:57:07 -!- Slereah has joined. 09:44:47 -!- tombom has joined. 09:51:53 -!- CakeProphet has joined. 09:52:51 -!- GreaseMonkey has quit ("X-Chat -> http://xchat.org <- At least when I quit I don't look like a lamer"). 10:43:32 -!- oerjan has joined. 11:04:25 -!- bsmntbombdood has quit (Read error: 110 (Connection timed out)). 11:30:37 -!- oerjan has quit ("leaving"). 11:56:29 -!- ineiros has quit (calvino.freenode.net irc.freenode.net). 11:56:33 -!- rodgort has quit (calvino.freenode.net irc.freenode.net). 11:56:33 -!- Ilari has quit (calvino.freenode.net irc.freenode.net). 11:56:37 -!- fungot has quit (calvino.freenode.net irc.freenode.net). 11:56:39 -!- fizzie has quit (calvino.freenode.net irc.freenode.net). 11:56:39 -!- Deewiant has quit (calvino.freenode.net irc.freenode.net). 11:56:39 -!- kerlo has quit (calvino.freenode.net irc.freenode.net). 11:56:42 -!- Slereah has quit (calvino.freenode.net irc.freenode.net). 11:56:43 -!- oklofok has quit (calvino.freenode.net irc.freenode.net). 11:56:44 -!- tombom has quit (calvino.freenode.net irc.freenode.net). 11:56:44 -!- Asztal_ has quit (calvino.freenode.net irc.freenode.net). 11:56:46 -!- neldoreth has quit (calvino.freenode.net irc.freenode.net). 11:56:47 -!- ehird has quit (calvino.freenode.net irc.freenode.net). 11:56:47 -!- GregorR has quit (calvino.freenode.net irc.freenode.net). 11:56:49 -!- Leonidas has quit (calvino.freenode.net irc.freenode.net). 11:56:49 -!- ski__ has quit (calvino.freenode.net irc.freenode.net). 11:56:51 -!- sebbu has quit (calvino.freenode.net irc.freenode.net). 11:56:52 -!- lament has quit (calvino.freenode.net irc.freenode.net). 11:56:52 -!- AnMaster has quit (calvino.freenode.net irc.freenode.net). 11:56:52 -!- mtve- has quit (calvino.freenode.net irc.freenode.net). 11:56:54 -!- dbc has quit (calvino.freenode.net irc.freenode.net). 11:56:56 -!- Dewi has quit (calvino.freenode.net irc.freenode.net). 11:56:56 -!- SimonRC has quit (calvino.freenode.net irc.freenode.net). 11:56:56 -!- CakeProphet has quit (calvino.freenode.net irc.freenode.net). 11:56:58 -!- Robdgreat has quit (calvino.freenode.net irc.freenode.net). 11:58:00 -!- CakeProphet has joined. 11:58:00 -!- tombom has joined. 11:58:00 -!- Slereah has joined. 11:58:00 -!- neldoreth has joined. 11:58:00 -!- oklofok has joined. 11:58:00 -!- Asztal_ has joined. 11:58:00 -!- ehird has joined. 11:58:00 -!- sebbu has joined. 11:58:00 -!- kerlo has joined. 11:58:00 -!- fungot has joined. 11:58:00 -!- mtve- has joined. 11:58:00 -!- fizzie has joined. 11:58:00 -!- ineiros has joined. 11:58:00 -!- Ilari has joined. 11:58:00 -!- rodgort has joined. 11:58:00 -!- GregorR has joined. 11:58:00 -!- Deewiant has joined. 11:58:01 -!- Dewi has joined. 11:58:01 -!- Robdgreat has joined. 11:58:01 -!- lament has joined. 11:58:01 -!- AnMaster has joined. 11:58:01 -!- Leonidas has joined. 11:58:01 -!- SimonRC has joined. 11:58:01 -!- dbc has joined. 11:58:01 -!- ski__ has joined. 14:45:32 -!- KingOfKarlsruhe has joined. 14:53:02 -!- Judofyr has joined. 15:29:38 -!- oklofok has quit (Read error: 110 (Connection timed out)). 15:33:54 -!- oklofok has joined. 15:50:21 -!- Slereah_ has joined. 16:00:16 -!- Hiato has joined. 16:01:11 -!- Slereah has quit (Read error: 110 (Connection timed out)). 16:16:50 -!- cherez1 has joined. 16:26:58 -!- tromp has joined. 16:32:33 -!- cherez1 has quit (Read error: 145 (Connection timed out)). 16:41:41 -!- cherez has joined. 16:52:14 -!- cherez has left (?). 17:04:15 Huh, PGP's source is publicly available. (Not FOSS, though.) 17:04:17 I didn't know. 17:09:53 -!- cherez has joined. 17:09:57 -!- cherez has left (?). 17:15:49 -!- Hiato has quit ("Leaving."). 17:23:24 -!- tombom has quit (Read error: 101 (Network is unreachable)). 17:52:34 -!- KingOfKarlsruhe has quit (Remote closed the connection). 17:54:03 ehird, heh 17:54:13 ehird, anything wrong with gpg though? 17:54:16 nope 17:55:14 there are even nice GUI frontends for gpg. And (this may surprise you) I use such a GUI frontend most of the time, because I find it hard to remember all the command line switches and what not 17:55:27 I know :P 17:55:29 ehird, anything like kgpg or such for OS X? 17:55:36 Yeah there's gpg stuff for os x 17:55:44 what I really want is something unifying gpg and ssh keys 17:55:56 ehird, like ssh-agent/gpg-agent? 17:56:03 or what do you mean 17:56:04 I mean the whole actual thing 17:56:09 Use your gpg identity as an ssh key 17:56:15 hm 17:56:15 instead of having two key pairs 17:56:32 I think they are technically incompatible though 17:56:41 you could make some combined file format or such I guess 17:56:53 Of course they're incompatible 17:56:56 I'm saying I wish they weren't :P 17:57:45 An irritating thing about gnupg is their refusal to provide a library 17:57:48 btw, do you use ssh-agent? 17:57:52 or gpg-agent 17:58:07 To be honest, I haven't got round to setting up an ssh key on my server yet. 17:58:12 heh 17:58:31 I'm still trying to think of a way to make sudo use my ssh key, thus completely removing passwords from the equation :-) 17:58:52 ehird, that is rather easy: generate a key, do: ssh-copy-id -i ~/.ssh/id_rsa.pub foo@bar.org 17:59:01 well you will need to adjust the paths of course 17:59:07 That's nothing to do with sudo 17:59:22 ehird, well true, but I meant copying the id is dead easy 17:59:31 Yes, I know. I've done it before :-) 17:59:34 AnMaster: Even easier: 17:59:40 just scp it to authorized_keys 17:59:47 .ssh/authorized_keys that is 18:00:07 4.16) Can't we have a gpg library? 18:00:07 This has been frequently requested. However, the current viewpoint of the GnuPG maintainers is that this would lead to several security issues and will therefore not be implemented in the foreseeable future. However, for some areas of application gpgme could do the trick. You'll find it at . 18:00:10 ^ don't like that attitude 18:00:41 ehird, what sort of security issues? Programmers using the library incorrectly? 18:00:47 I think that's the idea 18:00:49 but that's just stupid 18:00:50 mhm 18:01:01 you can use gpg's tools improperly too 18:01:12 if the API is high level enough it shouldn't be an issue 18:01:34 I mean, basically API for what you can do with the command line tool 18:01:41 yes 18:01:41 then it should be no more insecure 18:02:41 ehird, though since it is C I guess stuff like memory corruption in the program using the library could mean that a separate process would be more secure 18:02:57 AnMaster: hee you just reminded me of the debian issue 18:03:05 ehird, the openssl one? 18:03:07 yep 18:03:14 hur hur they're using uninitialized memory, well they must be stoopids! *comments out* 18:03:22 dumbo security guys ha ha 18:03:26 well that is the wrong way to solve it indeed. 18:03:43 I love how they did it just so valgrind would shut up 18:03:44 a better idea would have been to report it upstream 18:03:50 er 18:03:51 it wasn't a bug 18:03:57 it was an intentional source of entropy 18:03:58 mhm ok 18:04:09 ehird, not sure how random that would be 18:04:14 int main(void) { return 0; } /* 100% valgrind clean */ 18:04:37 AnMaster: in the process they commented out the main entropy source 18:04:39 ehird, only with the default suppressions for libc internals 18:04:51 so instead of just decreasing it a bit it pummeled it 18:04:52 and ld.so too 18:05:05 ehird, heh 18:05:17 ehird, I wonder what they did to python then.... 18:05:22 since it isn't valgrind clean 18:05:26 by design 18:05:26 :D 18:05:29 ln -s python perl 18:05:32 er 18:05:34 reverse that. 18:06:24 ehird, basically pythons' allocator does some quite (but not 100% in theory) safe stuff 18:06:29 err 18:06:32 python's* 18:06:36 quite safe stuff? 18:06:37 typo? :D 18:06:55 err? 18:07:14 ehird, what typo? 18:07:19 "basically python's allocator does some quite safe stuff" 18:07:20 the one I corrected? 18:07:23 doing safe stuff makes python complain now? 18:07:24 er 18:07:27 s/python/valgrind/ 18:07:39 ehird, no not exactly. 18:08:01 ehird, pymalloc reads uninitialised memory 18:08:19 ah. 18:08:20 let me check the details in README.valgrind in python's source tarball 18:08:20 why 18:09:14 PyMalloc needs to know whether an arbitrary address is one 18:09:14 that's managed by it, or is managed by the system malloc. 18:09:14 The current scheme allows this to be determined in constant 18:09:14 time, regardless of how many memory areas are under pymalloc's 18:09:14 control. 18:09:22 The memory pymalloc manages itself is in one or more "arenas", 18:09:22 each a large contiguous memory area obtained from malloc. 18:09:22 The base address of each arena is saved by pymalloc 18:09:23 in a vector. Each arena is carved into "pools", and a field at 18:09:25 the start of each pool contains the index of that pool's arena's 18:09:27 base address in that vector. 18:09:32 Given an arbitrary address, pymalloc computes the pool base 18:09:32 address corresponding to it, then looks at "the index" stored 18:09:33 near there. If the index read up is out of bounds for the 18:09:36 vector of arena base addresses pymalloc maintains, then 18:09:37 pymalloc knows for certain that this address is not under 18:09:39 pymalloc's control. Otherwise the index is in bounds, and 18:09:41 pymalloc compares 18:09:43 the arena base address stored at that index in the vector 18:09:45 to 18:09:47 the arbitrary address pymalloc is investigating 18:09:47 AnMaster: ... 18:09:48 dude. 18:09:49 that 18:09:51 is 18:09:53 some 18:09:54 fucking 18:09:56 flood 18:09:58 actually I should have pastebinned it yes 18:10:06 Python is so boringly conventional, its implementation does nooo fun tricks. 18:10:32 ehird, anyway point is that it reads uninitialised memory. 18:10:48 Rite. 18:12:10 -!- ais523 has joined. 18:12:26 hi ais523 18:12:36 ehird, strlen() could do that too, since the glibc optimised strlen() reads 4 or 8 bytes at a time to speed up, it means it could read past end of string a few bytes. But it is quite safe since it will read 4- or 8- aligned bytes. And the system pagesize is a multiple of that so it can never hit a non-readable page. 18:12:47 and hello ais523 18:12:47 ehird: I knew you'd said hi ais523 without even looking at the channel 18:12:47 your mom is so boringly conventional, she does nooo fun tricks 18:12:50 I know how glibc strlen() works, AnMaster. 18:12:54 just based on the fact I'd been highlighted 18:12:55 Because I advocated it. 18:12:56 and hi AnMaster 18:12:56 To comex. :P 18:13:08 ehird, anyway valgrind has a default suppression for that 18:13:26 AnMaster: what if there's a memory-mapped hardware address past the end of the string 18:13:37 and the string itself is entirely in memory-mapped hardware, but you don't mind reading those addresses? 18:14:22 ais523, hm, I don't think you can map anywhere but a multiple of the page size on x86 at least 18:14:27 About fun tricks (that we talked about here earlier): according to That One Wiki, people actually have been doing "swap palettes during the screen drawing to get more colors" tricks with CGA cards: "The best example of this in use is the game California Games[8] when run on a stock 4.77 MHz 8088. (Running it on a faster computer does not produce the effect, as the method the programmers used to switch palettes at predetermined locations is extremely sensitive 18:14:27 to machine speed.)" 18:14:34 and glibc have different strlen() for different arches 18:15:15 ehird, IMO they should valgrind should do: if (fileexists("/etc/debian-release")) disable_glibc_suppressions(); 18:15:17 ;) 18:15:29 then debian would surely get a superslow glibc 18:15:35 AnMaster: there are surely better ways to tell if you're on Debian or not! 18:15:47 and why would you deliberately give debian a superslow glibc? 18:16:01 AnMaster: eh, debian is alright; just a fuckup because they modified security software without carefully checking it (which _is_ stupid) 18:16:10 ais523, Well since we discussed Debian, OpenSSL and valgrind... 18:16:56 "relying on the garbage collector is a crutch" /facepalm 18:17:07 heh, who wrote that? 18:17:12 random redditor 18:17:21 it's a fairly wide-spread view 18:17:34 lament: yeah, like AnMaster here used to have ;) 18:18:04 i'm not saying people who think that aren't insane, but there's logic to their madness 18:18:18 er 18:18:20 there is? 18:18:20 ehird, btw using unintialised memory for entropy isn't such a good idea, since kernel usually give blanked out pages to glibc, so the stuff you end up reading is likely your own previous stuff. 18:18:54 I'm not sure about memory allocated with sbrk() but not blanking pages before giving them to user space sounds like a really bad idea for security 18:21:02 * ehird works on Awful Project 18:21:47 ehird, about GC. GC makes sense for languages with typed data, GC for C programs is IMO not such a good idea. Well good idea maybe, but in practise it doesn't work well. 18:21:55 Well,* 18:22:15 That's a C deficiency 18:22:21 ehird, indeed. 18:22:41 sea deficiency :( 18:22:52 ;__; 18:22:55 lament, that would be a boat or a whale? 18:23:17 vitamin C deficiency? 18:23:22 haha 18:24:09 According to http://wiki.debian.org/SSLkeys#TechnicalSummary it's not that they were using unitialized memory for randomness; it's just that they removed two MD_Update calls, of which one actually was responsible of adding /dev/random-based entropy in. 18:24:30 anyway... since C and C++ are quite popular languages even for stuff that don't need the low levelness or the speed. it isn't strange that GCs ended up with a bad reputation(sp?) 18:24:31 hmm 18:24:36 fizzie: yes. I said that 18:24:55 ehird: There was too much babble for me to read through that closely. 18:24:57 AnMaster: it's strange that C is popular for anything but kernel development 18:25:06 well that's strange too but I won't be too heretical 18:25:11 ehird, well C++ is sadly more popular these days 18:25:21 it's strange that foo is popular for bar 18:25:26 'This was done to try to make it easier to debug C applications that use the openssl libraries which is a good thing to do." 18:25:33 Is this the Simple English debian wiki? 18:25:51 ehird: you're the only person on *the internet* who complains more than I do. 18:26:13 lament: Oh, no, I know one person who complains more. 18:26:20 wow. 18:26:23 lament: well, you don't complain much in here 18:26:26 lament: They are a huge pain to talk to. 18:26:31 but presumably based on that you complain a lot elsewhere 18:26:42 ais523: he complains loads in here 18:26:45 ehird, on the other hand since OS is in C, basic libraries in C and graphical window drawing thingy is either built into kernel or a program coded in C (all of those can be justified to be in C, they need low level stuff and they need speed), it is easy to interface with stuff if you code in C 18:26:58 and then other stuff can interface with the stuff you just wrote in C as easily 18:26:58