| --- Log opened Wed Sep 17 00:00:11 2014 |
| 00:03 | <@gnolam> | Oh hey, a 32-bit float seems to make it through float(str()) unscathed. That simplifies things. |
| 00:03 | <&McMartin> | If it's super-important that it do so you may be able to get it to dump in binary/hex floating point format |
| 00:11 | | froztbyte [froztbyte@Nightstar-frrora.za.net] has quit [Operation timed out] |
| 00:11 | | froztbyte [froztbyte@Nightstar-frrora.za.net] has joined #code |
| 00:11 | | mode/#code [+o froztbyte] by ChanServ |
| 00:23 | <&ToxicFrog> | Reiv: "mostly the same" is honestly worse than "totally different" |
| 00:32 | <@Reiv> | ToxicFrog: I agree on that |
| 00:32 | <@RchrdB> | gnolam, have you tested that for all 4-billion-ish possible values? :) |
| 00:32 | <@Reiv> | Though it does at least make translation more straightforward (mostly) |
| 00:42 | <@gnolam> | Nope. |
| 00:43 | <@TheWatcher> | It just means you think you've got it right, only you haven't in subtle and horrible ways |
| 00:43 | <@TheWatcher> | Generally involving spiders. |
| 00:44 | <@gnolam> | But it seems to be true for enough tested values that any deviation will be well within the error introduced by the subsequent calculations done on said values. |
| 00:46 | <@gnolam> | (And the errors introduced when the calculated values meet the real world and the imperfect machines they control) |
| 00:46 | <@Reiv> | oh hush |
| 00:46 | <@Reiv> | I spent my first three months at this job converting TSQL to PLSQL |
| 00:47 | <@gnolam> | I may well write a unit test just for my peace of mind, but. It'll be good enough. |
| 00:47 | <@Reiv> | And aside from unfortunate bits involving date functions or temporary tables, the conversion process was straightforward enough |
| 00:47 | <@Reiv> | Indeed, large swathes of it could well have been done automatically, as the dialects are close enough |
| 00:48 | <@RchrdB> | gnolam, I'm semi-serious about that though; one of the nice things about 32-bit floats is that there are "only" about 4 billion of them, so you can with modern CPU speeds around 2-3GHz *exhaustively* test 1-input float functions at roughly 1 second-ish per (cycle per call). :) |
| 00:48 | <@TheWatcher> | I now invite you to write a system that works with mysql, postgres, and oracle |
| 00:49 | <@TheWatcher> | You basically end up having to write a database abstraction layer to avoid the horrible crawling madness |
| 00:50 | <@Reiv> | TheWatcher: On that point I do agree |
| 00:50 | | RchrdB [0x2ba22e11@Nightstar-qe9.aug.187.81.IP] has quit [[NS] Quit: Leaving] |
| 00:50 | <@Reiv> | But, uh, we have a BI system |
| 00:50 | <@Reiv> | For that exact reason |
| 00:51 | <@Reiv> | The funiest bit of the whole damn thing was the day that our analysts discovered that after all the effort of SQL-ising their queries, they could just upload bloody excel files as a data source. |
| 00:51 | <@Reiv> | We, uh, have to watch that like a dragon on his hoard~ |
| 01:02 | | Checkmate [Z@Nightstar-484uip.cust.comxnet.dk] has quit [Ping timeout: 121 seconds] |
| 01:17 | <@Azash> | I'm usually torn |
| 01:17 | <@Azash> | I work with Rails, so I use ActiveRecord |
| 01:17 | <@Azash> | Which is really nice and pleasant |
| 01:17 | <@Azash> | And then you have the console up and you see it handle a request and suddenly there's like four empty transactions |
| 01:17 | <@Azash> | Just BEGIN COMMIT BEGIN COMMIT BEGIN COMMIT BEGIN COMMIT |
| 01:31 | | * iospace commits Azash to the asylum |
| 01:31 | <@iospace> | :P |
| 01:37 | | * Reiv begins to commit Azash to the asylum; get it right~ |
| 01:59 | <~Vornicus> | RchrdB: I found myself in need of testing trios of floats a couple weeks back |
| 02:00 | <@gnolam> | He's gone. |
| 02:00 | <~Vornicus> | bah |
| 02:05 | <@gnolam> | FOREIGN KEY (sample) REFERENCES Xen(portal) ON RESONANCE CASCADE |
| 02:11 | <&McMartin> | The right query in the wrong place can make *all the difference in the world* |
| 02:16 | | * Vornicus spent his afternoon crafting queries |
| 02:18 | <~Vornicus> | I'm still dissatisfied with my result for the "use a single query, get counts of children from two different tables" |
| 02:20 | | VirusJTG [VirusJTG@Nightstar-6i5vf7.sta.comporium.net] has quit [[NS] Quit: Program Shutting down] |
| 02:41 | <@Reiv> | Why are children in two different tables |
| 02:42 | <~Vornicus> | because they're two different types of children |
| 02:42 | <@Reiv> | I take it we are not talking 'boys' and 'girls' then |
| 02:42 | <~Vornicus> | I described the situation earlier - posts, comments, votes |
| 02:43 | <@Reiv> | oh right |
| 02:43 | <@Reiv> | Yes that's fair |
| 02:43 | <@Reiv> | Why the dissatisaction? Wrong number? |
| 02:43 | <~Vornicus> | Number of rows generated en route to the final result |
| 02:45 | <~Vornicus> | If a post has 3 votes and 5 comments, my double-left-join thing gets (as far as I can tell, unless my system understands this situation and optimizes accordingly) 15 rows for that one post, as opposed to 3 (if I just counted votes) or 5 (if I just counted comments) |
| 02:48 | <~Vornicus> | -- it goes basically select post_id, count(distinct comment_id), count(distinct vote_id) from posts left join comments left join votes |
| 02:49 | <@Reiv> | And you'd want it to get 8, correct? |
| 02:49 | <~Vornicus> | yeah, or something |
| 02:50 | <@Reiv> | Yeah, you've got a misjoin, that's always fun |
| 02:50 | <@Reiv> | Er, crossjoin |
| 02:50 | | * Reiv muses |
| 02:50 | <~Vornicus> | especially as this gets quadratically worse the more popular a single post (or whatever) gets |
| 02:50 | <~Vornicus> | (there are of course conditions on the two joins but they're cross to each other) |
| 02:52 | <@Reiv> | I've solved this one before |
| 02:52 | <@Reiv> | what language you typing in? |
| 02:53 | <~Vornicus> | It's MySQL running behind PHP/CodeIgniter; this and similar monsters are in views. |
| 02:53 | <@Reiv> | hm |
| 02:53 | <@Reiv> | Do you have the UNION command available? |
| 02:53 | <~Vornicus> | Yepperoni. |
| 02:54 | <~Vornicus> | Oh, I see where you're going. Subquery and then do counts. |
| 02:54 | <@Reiv> | Because my default solution would be SELECT (counts and shit) FROM (vote_id_query UNION comment_id_query) |
| 02:54 | <@Reiv> | So long as those two queries return the same column names, you're golden |
| 02:54 | <@Reiv> | And maintainable, to boot |
| 02:54 | <~Vornicus> | Indeed, that looks very sensible. |
| 02:55 | <@Reiv> | Table-specific rules go in their individual subqueries, generic rules go in the wrapper query |
| 02:55 | | * Reiv crackles with power! |
| 02:55 | <@Reiv> | It's also totally doable with joins, but then you're rocking a crossjoin and null-rules, and meh |
| 02:55 | <@Reiv> | Save the WHERE foo_id IS NULL for where you need it~ |
| 02:56 | | macdjord [macdjord@Nightstar-7rac1r.mc.videotron.ca] has quit [Ping timeout: 121 seconds] |
| 02:57 | <@Reiv> | Do make a point of poking me if you have other SQL logic you need a hand with, eh? |
| 02:57 | <~Vornicus> | Of course. Though usually I'm fighting with this before you get up in the morning, I think~ |
| 02:57 | <@Reiv> | I don't check the channel all that much. |
| 02:57 | <@Reiv> | haha |
| 02:58 | <~Vornicus> | I will give you a ping when I need it, though, thank you |
| 02:58 | <@Reiv> | no worries |
| 02:58 | <@Reiv> | You've helped me often enough~ |
| 02:58 | <@Reiv> | agh |
| 02:58 | <@Reiv> | You've helped me often enough, it's nice to be able to return the favor~ |
| 03:00 | | * Reiv muses |
| 03:00 | <@Reiv> | I am generally hear, uh, five hours ago |
| 03:00 | <@Reiv> | *here |
| 03:00 | <@Reiv> | What time is that local for you these days, you moved |
| 03:00 | <~Vornicus> | yeah that's when I get out of work. 5PM |
| 03:01 | <@Reiv> | ha, so |
| 03:01 | <@Reiv> | If you've got anything left bugging you when you leave work, I can probably help~ |
| 03:01 | <@Reiv> | That's kind of amusingly fitting. |
| 03:02 | <@Reiv> | Yeah, crossjoins are a pain in the butt |
| 03:04 | <@Reiv> | tsk, though you lack analytic functions, oh well |
| 03:04 | <@Reiv> | Not the end of the world but they are my Sharp Knife in oh so many sqlcancers~ |
| 03:05 | | macdjord [macdjord@Nightstar-7rac1r.mc.videotron.ca] has joined #code |
| 03:05 | | mode/#code [+o macdjord] by ChanServ |
| 03:22 | | gnolam [lenin@Nightstar-sanfmp.cust.bredbandsbolaget.se] has quit [Ping timeout: 121 seconds] |
| 03:48 | | gnolam [lenin@Nightstar-sanfmp.cust.bredbandsbolaget.se] has joined #code |
| 03:48 | | mode/#code [+o gnolam] by ChanServ |
| 04:09 | <@Alek> | http://imgur.com/gallery/viQOQgO |
| 04:48 | | Kindamoody[zZz] is now known as Kindamoody |
| 04:52 | | Vornicus [vorn@ServerAdministrator.Nightstar.Net] has quit [[NS] Quit: Leaving] |
| 05:08 | | Harlow [harlow@Nightstar-pq0497.il.comcast.net] has joined #code |
| 05:38 | | Turaiel is now known as Turaiel[Offline] |
| 06:19 | | Harlow [harlow@Nightstar-pq0497.il.comcast.net] has quit [[NS] Quit: BED] |
| 06:46 | | celticminstrel [celticminst@Nightstar-ak6p6n.dsl.bell.ca] has quit [[NS] Quit: And lo! The computer falls into a deep sleep, to awake again some other day!] |
| 08:19 | | Kindamoody is now known as Kindamoody|out |
| 08:47 | | [R] [rstamer@genoce.org] has quit [Ping timeout: 121 seconds] |
| 09:35 | | [R] [rstamer@Nightstar-d7h8ki.org] has joined #code |
| 10:17 | | RchrdB [0x2ba22e11@Nightstar-952.jvr.168.194.IP] has joined #code |
| 10:17 | | mode/#code [+o RchrdB] by ChanServ |
| 10:39 | | RchrdB [0x2ba22e11@Nightstar-952.jvr.168.194.IP] has quit [Operation timed out] |
| 10:44 | | RchrdB [0x2ba22e11@Nightstar-952.jvr.168.194.IP] has joined #code |
| 10:44 | | mode/#code [+o RchrdB] by ChanServ |
| 11:00 | | * Tarinaky head-desks at her own stupidity. |
| 11:03 | | Checkmate [Z@Nightstar-484uip.cust.comxnet.dk] has joined #code |
| 11:03 | | mode/#code [+o Checkmate] by ChanServ |
| 11:14 | | VirusJTG [VirusJTG@Nightstar-6i5vf7.sta.comporium.net] has joined #code |
| 12:11 | <@Tarinaky> | I am an even bigger idiot. |
| 12:12 | <@Tamber> | How badly has it caught fire? |
| 12:16 | <@Tarinaky> | I was freeing something on the stack, stupidly... |
| 12:17 | <@Tarinaky> | SO I got rid of the free statement, pushed it back through valgrind and went to get food. |
| 12:17 | <@RchrdB> | forgot to recompile? |
| 12:17 | <@Tarinaky> | Then on the way there I realised that I was freeing something on the stack because I shoul;d have been /copying/ the damn thing onto the heap. |
| 12:17 | <@Tarinaky> | So I've just been filling up a bunch of structs with identical values! |
| 12:24 | <@Tarinaky> | Invalid read of size 1 |
| 12:24 | <@Tarinaky> | ==4266== at 0x4836484: strlen (mc_replace_strmem.c:390) |
| 12:24 | <@Tarinaky> | I have no idea what this means :/ |
| 12:26 | <@TheWatcher> | you're trying to access memory you don't have access to. |
| 12:27 | <@TheWatcher> | either via a pointer to a junk location, or to memory you've freed already |
| 12:27 | <@Tarinaky> | Google says the issue is that I need to calloc one more element than strlen ( source ) |
| 12:28 | <@TheWatcher> | This is true |
| 12:28 | <@Tarinaky> | (Yes, I suck at pure-C. I'm a C++ coder. Go figure) |
| 12:28 | <@Tarinaky> | Inb4 C++ coder == terrible coder |
| 12:29 | <@TheWatcher> | printf("%d\n", strlen("foo")); will print out '3', but the string is actually 4 characters long because it needs to have the '\0' on the end. |
| 12:29 | <@Tarinaky> | Yes, I'd forgotten this. |
| 12:30 | <@Tarinaky> | I was expecting strlen to include the \0 |
| 12:30 | <@TheWatcher> | That would make too much sense! |
| 12:33 | <@Tarinaky> | Well, that's a good sign. Valgrind doesn't immediately shout at me. |
| 12:34 | <@TheWatcher> | always good, that. |
| 12:36 | <@Tarinaky> | SHit, no. Infiniloop :/ |
| 12:37 | | * Tarinaky facepalms. |
| 12:50 | | * TheWatcher eyes this |
| 12:50 | <@TheWatcher> | Oh gods, I'm going to have to generate a fucking dot file for this |
| 12:59 | <@RchrdB> | as in graphviz? graphviz is pretty civilised |
| 13:00 | <@Tarinaky> | I think he means /home/foo/.program |
| 13:04 | < Syka> | psst |
| 13:04 | < Syka> | http://labs.twistedmatrix.com/2014/09/twisted-1401-released.html |
| 13:06 | <@Tarinaky> | Are source and destination allowed to overlap for strcpy? |
| 13:06 | <@Tarinaky> | If I want to implement a byte-shift left. |
| 13:06 | <@Tarinaky> | The source is 'ahead' of the destination by a character. |
| 13:06 | <@Tarinaky> | Or is this still a shooting offence? |
| 13:07 | <@TheWatcher> | RchrdB: yes, graphviz |
| 13:08 | <@TheWatcher> | for a graph that is currently looking like having about 8000 nodes |
| 13:10 | <@TheWatcher> | Tarinaky: generally no |
| 13:10 | <@Tarinaky> | What about in this specific case? |
| 13:10 | <@RchrdB> | TheWatcher, ehhhhhh~ |
| 13:11 | <@RchrdB> | Like I say, dot is a pretty civilised text format. |
| 13:11 | <@TheWatcher> | strcpy() is often implmented either with memcpy (which explicitly states areas should not overlap) or with a loop that copies bytes sequentially |
| 13:11 | <@RchrdB> | IME, the hard part is repeatedly changing the code that generates your dotfiles until you find a rule for computing a subset of the graph's nodes and edges that makes the whole thing comprehensible to look at when rendered. |
| 13:12 | <@TheWatcher> | so you'd obliterate your \0, and run off into bufferoverlowland |
| 13:13 | <@RchrdB> | Tarinaky, memmove() is the function to use if you want src and dest areas that may overlap. |
| 13:13 | <@RchrdB> | strcpy()'s linux man page says that the strings passed to strcpy may not overlap. |
| 13:14 | <@RchrdB> | also strcpy() is a slightly-evil function, strncpy() slightly less so. |
| 13:14 | <@TheWatcher> | (although beware strncpy()'s hilarious gotcha with not nul-terminating |
| 13:14 | <@TheWatcher> | ) |
| 13:15 | <@RchrdB> | TheWatcher, e.g. last time I was using graphviz in anger, I was drawing graphs of "referrer â URL" in order to work out why a piece of routing logic had given a user a link to a bad page, and I had to make a couple of runs at things like pruning the CSS files off the graph, since they were referred to by EVERY other node and made it unreadable. :) |
| 13:17 | <@TheWatcher> | I'm trying to visuals a system that has an N-way tree of 'metadata' nodes, each of those nodes can have zero or more entities associated with it that use the metadata to determine roles, and zero or more role/user associations |
| 13:18 | | * Tarinaky comes up with a more robust way of doing what she wanted using strtok |
| 13:22 | <@Tarinaky> | Wait. What. |
| 13:25 | <@Tarinaky> | Oh. |
| 13:26 | <@Tarinaky> | Not to self. strtok "\"foo\"" -> [ "foo" ] /not/ [ "", "foo", "" ] |
| 13:26 | <@Tarinaky> | Err, with "\"" as the seperator |
| 13:44 | <&ToxicFrog> | strtok :gonk: |
| 13:47 | <@Tarinaky> | ? |
| 13:58 | <&ToxicFrog> | ":gonk:" is an expression of horror and dismay. |
| 13:59 | <@Tarinaky> | Why? |
| 14:08 | <@Tarinaky> | I mean, why're you horrified? |
| 14:08 | <@Tarinaky> | (I'm using a CSV file format specifically so I can parse it without extra includes) |
| 14:14 | <@Tarinaky> | The take home lesson from this is that I suck at writing pure C and should never do this for a living; I think~ |
| 14:17 | | Turaiel[Offline] is now known as Turaiel |
| 14:17 | <&ToxicFrog> | It destructively modifies its argument in place and is not thread- or even interleave-safe. |
| 14:21 | <@Tarinaky> | Well, yes. |
| 14:22 | <@Tarinaky> | I thought you were :gonk: ing at me for using it. |
| 14:22 | <@Tarinaky> | Not for it being slightly... dated? |
| 14:23 | <&ToxicFrog> | Well, yes, I was, because it is the gets() of string processing |
| 14:34 | | Alek [omegaboot@Nightstar-c8t.a00.36.73.IP] has quit [Ping timeout: 121 seconds] |
| 14:35 | | Checkmate [Z@Nightstar-484uip.cust.comxnet.dk] has quit [[NS] Quit: If I had a world of my own, everything would be nonsense. Nothing would be what it is because everything would be what it isn't. And contrary-wise; what it is it wouldn't be, and what it wouldn't be, it would. You see?] |
| 14:37 | | Alek [omegaboot@Nightstar-c8t.a00.36.73.IP] has joined #code |
| 14:37 | | mode/#code [+o Alek] by ChanServ |
| 14:37 | | Turaiel is now known as Turaiel[Offline] |
| 15:42 | <@Tarinaky> | ToxicFrog: It's also the easiest way to parse a CSV file in pure-C |
| 16:01 | | Attilla [uid13723@Nightstar-ed0oqj.irccloud.com] has joined #code |
| 16:18 | | celticminstrel [celticminst@Nightstar-ak6p6n.dsl.bell.ca] has joined #code |
| 16:18 | | mode/#code [+o celticminstrel] by ChanServ |
| 16:30 | | * Tarinaky grumbles, unable to find the dilbert she's looking for. |
| 16:30 | <@Tarinaky> | It's onw where Asok is saying that they've completed everything |
| 16:34 | <@Tarinaky> | Ahah! |
| 17:16 | <@RchrdB> | Tarinaky, would you like a BSD-licensed CSV parser? I can probably write a finite state machine for it. :P |
| 17:17 | <@RchrdB> | What output format would you like? |
| 17:32 | | celticminstrel [celticminst@Nightstar-ak6p6n.dsl.bell.ca] has quit [Connection closed] |
| 17:32 | | celticminstrel [celticminst@Nightstar-ak6p6n.dsl.bell.ca] has joined #code |
| 17:32 | | mode/#code [+o celticminstrel] by ChanServ |
| 17:34 | | thalass [thalass@Nightstar-up51sl.bigpond.net.au] has quit [Ping timeout: 121 seconds] |
| 17:41 | <@Tarinaky> | RchrdB: "No thanks." |
| 17:41 | <@Tarinaky> | My program is done |
| 17:41 | <@Tarinaky> | And while it counts as premature optimisation: but binary size matters in context. |
| 17:43 | <@RchrdB> | I'd be willing to bet I could write a function that breaks an in-memory CSV function into lines in a few kB. :P |
| 17:44 | <@Tarinaky> | That's require effort on my part. |
| 17:46 | | thalass [thalass@Nightstar-up51sl.bigpond.net.au] has joined #code |
| 17:46 | | mode/#code [+o thalass] by ChanServ |
| 17:57 | | Turaiel[Offline] is now known as Turaiel |
| 18:46 | | Turaiel is now known as Turaiel[Offline] |
| 18:48 | | RchrdB [0x2ba22e11@Nightstar-952.jvr.168.194.IP] has quit [Operation timed out] |
| 19:38 | | Kindamoody|out is now known as Kindamoody |
| 19:56 | <@froztbyte> | https://github.com/froztbyte/pushloop if anyone ever needs that kind of terrible |
| 19:56 | <@froztbyte> | "I got tired of running this command myself" is the design motivation |
| 19:57 | <@froztbyte> | (and the screen bit for being visible) |
| 19:57 | <@Tamber> | :) |
| 19:57 | <@froztbyte> | mostly for when you're doing supercheap stuff in a shared screen session or somesuch |
| 20:15 | | Kindamoody is now known as Kindamoody[zZz] |
| 20:51 | | gnolam_ [lenin@Nightstar-sanfmp.cust.bredbandsbolaget.se] has joined #code |
| 20:55 | | gnolam [lenin@Nightstar-sanfmp.cust.bredbandsbolaget.se] has quit [Ping timeout: 121 seconds] |
| 21:02 | | gnolam_ is now known as gnolam |
| 21:02 | | mode/#code [+o gnolam] by ChanServ |
| 21:02 | | thalass [thalass@Nightstar-up51sl.bigpond.net.au] has quit [Ping timeout: 121 seconds] |
| 21:15 | | thalass [thalass@Nightstar-l4g186.bigpond.net.au] has joined #code |
| 21:15 | | mode/#code [+o thalass] by ChanServ |
| 21:22 | | thalass [thalass@Nightstar-l4g186.bigpond.net.au] has quit [Ping timeout: 121 seconds] |
| 21:35 | | thalass [thalass@Nightstar-nad7or.bigpond.net.au] has joined #code |
| 21:35 | | mode/#code [+o thalass] by ChanServ |
| 21:54 | | thalass [thalass@Nightstar-nad7or.bigpond.net.au] has quit [Ping timeout: 121 seconds] |
| 22:07 | | thalass [thalass@Nightstar-7cct2u.bigpond.net.au] has joined #code |
| 22:07 | | mode/#code [+o thalass] by ChanServ |
| 22:15 | | * TheWatcher sighs at code design |
| 22:30 | <&McMartin> | ;_; |
| 22:30 | | * McMartin is worried about that right now too |
| 22:31 | <@TheWatcher> | Thankfully this isn't something I need to so much /worry/ about, as just get kinda irritated if I get it wrong. |
| 22:34 | <@TheWatcher> | I think I'm going to have to muck around with template specialisation, joy. |
| 22:35 | <&McMartin> | Oh man. |
| 22:35 | <&McMartin> | Good luck. -_- |
| 22:35 | <&McMartin> | ... Oh wait, I can have one minor actually useful piece of advice |
| 22:35 | | thalass [thalass@Nightstar-7cct2u.bigpond.net.au] has quit [Ping timeout: 121 seconds] |
| 22:36 | <@TheWatcher> | Oh? |
| 22:36 | <&McMartin> | Try to snag access to a machine with clang on it; clang's template errors are often (not a slam dunk, but often) more human-useful than g++'s. |
| 22:37 | | Checkmate [Z@Nightstar-484uip.cust.comxnet.dk] has joined #code |
| 22:37 | | mode/#code [+o Checkmate] by ChanServ |
| 22:40 | <@TheWatcher> | Thanks, I should be able to arrange that easily enough; I can probably stick it on morpheus |
| 22:48 | | thalass [thalass@Nightstar-gf3dmq.bigpond.net.au] has joined #code |
| 22:48 | | mode/#code [+o thalass] by ChanServ |
| --- Log closed Thu Sep 18 00:00:27 2014 |