| --- Log opened Thu Feb 17 00:00:02 2011 |
| 00:01 | < RichardBarrell> | Oops, I think that makes more sense if you swap "sentences" and "postfix" around. |
| 00:04 | | * Reiv wrangles a malformed Excel expression. |
| 00:04 | < Reiv> | Stupid nested if statements. |
| 00:04 | < Reiv> | =IF(ISTEXT(L2),IF(L2="Other","Other",CONCATENATE("to ", L2, " by ", IF(O2="N/A","Other",O2))," ")) |
| 00:05 | < Reiv> | Too many arguements. |
| 00:06 | < Reiv> | This frankenstein is basically going: If L2 is empty, foo is " ". If not, if L2 has the text "Other", foo is "Other." If not, concatenate L2 and O2 together. If O2 is "N/A", concatenate it as if O2 had the text "Other". |
| 00:08 | < Reiv> | Intended results are basically "To X by Y". If X is Other, don't bother just record it all as "Other". If Y is Other or N/A, record it as "To X by Other." |
| 00:14 | <@ToxicFrog> | Brb, getting bucket |
| 00:14 | | * ToxicFrog vomits black, acidic bile from his eye sockets |
| 00:15 | <@McMartin> | FACT: Men love spreadsheets for their wide array of industrial applications |
| 00:15 | <@McMartin> | FACT: Four out of five men are crushed to death by reference manuals every day |
| 00:15 | | AnnoDomini [annodomini@Nightstar-ae40b7e9.adsl.tpnet.pl] has quit [[NS] Quit: leaving] |
| 00:20 | < RichardBarrell> | McMartin: I have a copy of the O'Reilly's "Java 5 in a Nutshell" and I could hospitalise somebody with it if I applied sufficient effort and creativity. It's that heavy. :) |
| 00:21 | < Alek> | you could hospitalize somebody with a rolled-up newsweek. that's no judge of weight. |
| 00:22 | < Reiv> | McMartin: Yeah, it's starting to feel like that :( |
| 00:25 | < RichardBarrell> | Alek: true, but only one. You could ration this Java book out in pieces and take out an entire family. |
| 00:26 | <@McMartin> | vprintf("You dawg I heard you like %s so I put a %s in your %s so you can %s while you %s\n", va); |
| 00:26 | <@McMartin> | s/You/Yo/ |
| 00:26 | | Gruber [lenin@Nightstar-38637aa0.priv.bahnhof.se] has joined #code |
| 00:26 | <@ToxicFrog> | Somehow I'm not surprised that even the "nutshell" book on Java is hefty enough to kill~ |
| 00:27 | < simon_> | somehow, writing a thick book about Java called "... in a nutshell" seems like raping the metaphor. |
| 00:27 | | gnolam [lenin@Nightstar-38637aa0.priv.bahnhof.se] has quit [Ping timeout: 121 seconds] |
| 00:28 | < RichardBarrell> | simon_: seemed like sophisticated humour to me. :) |
| 00:28 | | You're now known as TheWatcher[T-2] |
| 00:30 | | You're now known as TheWatcher[zZzZ] |
| 00:32 | < RichardBarrell> | McMartin: I see your vprintf and I raise you printf("Yo dawg, I heard you like %1$s so I put a %1$s in your %1$s so you can %1$s while you %1$s.\n", noun); /* because nobody should have to put up with va_list. */ |
| 00:33 | < simon_> | RichardBarrell, maybe it was an oversize coconut. |
| 00:34 | <@McMartin> | There's no guarantee they're all the same noun, or that the verb is the noun |
| 00:35 | < RichardBarrell> | McMartin: I'm aware of the limitations, but it's the first place I've come across yet where printf's support for positional specifiers seemed like it might actually be useful for something. |
| 00:35 | | Gruber is now known as gnolam |
| 00:36 | < RichardBarrell> | For dubious and stupid values of the word 'useful'. |
| 00:41 | < simon_> | fgets(buf, BUFLEN, stdin); gives me: |
| 00:41 | < simon_> | warning: passing argument 1 of 'fgets' discards qualifiers from pointer target type |
| 00:41 | < simon_> | /usr/include/stdio.h:604: note: expected 'char * __restrict__' but argument is of type 'const char *' |
| 00:43 | < RichardBarrell> | simon_: what's buf declared as? The __restrict__ thing is almost a red herring here, it just means that fgets is allowed to shit on your cornflakes if, say, "buf" and "stdin" turn out to be pointers to the same piece of memory. |
| 00:43 | < simon_> | RichardBarrell, as const char buf[BUFLEN]. |
| 00:43 | < RichardBarrell> | (Not that the type aliasing rules don't ban that anyway.) |
| 00:44 | < RichardBarrell> | simon_: right, so there's your problem. fgets expects its first parameter to be a (char*), rather than a (const char*), because fgets plans to modify the contents of buf. |
| 00:44 | < simon_> | RichardBarrell, I added the const because of the man page to see if the warning would go away. |
| 00:44 | | TarinakyKai [Tarinaky@Nightstar-f349ca6d.plus.com] has quit [Connection closed] |
| 00:44 | < RichardBarrell> | You still get that warning without the "const"? |
| 00:45 | < simon_> | d'oh. |
| 00:45 | < simon_> | let me try. |
| 00:45 | < simon_> | no :) I was mixing the man pages between fgets and sscanf ;P |
| 00:46 | < RichardBarrell> | Happens to us all. ^^ |
| 00:47 | < simon_> | thanks. |
| 00:48 | < RichardBarrell> | You mark a variable with (const) to declare that you don't plan to modify it - so sscanf takes a (const char*) rather than a (char*) because it never alters the chars that the pointer leads to. |
| 00:48 | < RichardBarrell> | No trouble at all.m |
| 00:48 | < RichardBarrell> | Perhaps you could, in return, teach me to type? |
| 00:48 | | * RichardBarrell headdesks and s/all\.m$/all./ |
| 00:52 | < Reiv> | hahaha, was missing a bracket |
| 01:10 | | gnolam [lenin@Nightstar-38637aa0.priv.bahnhof.se] has quit [[NS] Quit: Z?] |
| 01:27 | | Derakon[AFK] is now known as Derakon |
| 01:34 | <@Vornicus-Latens> | Well, anyway, this is an improvement. http://funkyhorror.net/vornicus/multiplication_drill.html |
| 01:39 | <@Vornicus-Latens> | (added instructions) |
| 01:41 | | Vornicus-Latens is now known as Vornicus |
| 01:47 | | SK [smith@DCDEB4.F95CFD.2EEAA4.B1AE3D] has joined #code |
| 01:47 | | SmithKurosaki [smith@Nightstar-7820a96a.dsl.teksavvy.com] has quit [Connection reset by peer] |
| 01:51 | | Derakon [Derakon@Nightstar-cfae48c3.ca.comcast.net] has quit [Ping timeout: 121 seconds] |
| 02:50 | | SK is now known as SmithKurosaki |
| 02:54 | | cpux is now known as shade_of_cpux |
| 03:21 | | celticminstrel [celticminst@Nightstar-f8b608eb.cable.rogers.com] has joined #code |
| 03:56 | | Kindamoody is now known as Kindamoody[zZz] |
| 04:11 | < simon_> | I wonder, if various general-purpose malloc implementations are superior to glibc's, why isn't glibc's replaced with one of these? |
| 04:15 | <@McMartin> | I was unaware there remained serious problems with glibc's. |
| 04:22 | < simon_> | oh man, I've just been trolled by jemalloc's author then. |
| 04:23 | < simon_> | or rather, he was a little sales pitchy. but he showed some benchmarks and I took them for granted. they apparently showed that tcmalloc and jemalloc perform best among the ones he tried. |
| 04:24 | <@Vornicus> | yeah, I never heard of actual problems with glibc's malloc. |
| 04:25 | <@McMartin> | The "serious" there is important |
| 04:25 | <@McMartin> | It's entirely plausible that on certain kinds of workloads other allocators will outperform it |
| 04:29 | <@Vornicus> | Whether these "certain kinds of workloads" are actually the common case, I don't know. |
| 05:16 | | celticminstrel [celticminst@Nightstar-f8b608eb.cable.rogers.com] has quit [[NS] Quit: And lo! The computer falls into a deep sleep, to awake again some other day!] |
| 06:21 | < Alek> | "I work in an electronics store. An obviously affluent woman comes in, browses for a while, and buys a $20 digital photo frame. On her way out, she pauses by a holster display, digs out her phone, and calls someone, presumably her husband. 'All done, I bought Xander an iPad for his graduation gift.'" |
| 07:24 | | Kindamoody[zZz] is now known as Kindamoody |
| 07:39 | | AnnoDomini [annodomini@Nightstar-ae40b7e9.adsl.tpnet.pl] has joined #code |
| 07:39 | | mode/#code [+o AnnoDomini] by Reiver |
| 08:06 | | Vornicus is now known as Vornicus-Latens |
| 08:23 | | Kindamoody is now known as Kindamoody|out |
| 08:58 | | You're now known as TheWatcher |
| 09:51 | | RichardBarrell [mycatverbs@Nightstar-f68eb197.cable.virginmedia.com] has quit [Ping timeout: 121 seconds] |
| 10:00 | | AnnoDomini [annodomini@Nightstar-ae40b7e9.adsl.tpnet.pl] has quit [[NS] Quit: I see.] |
| 10:35 | | AnnoDomini [annodomini@F67919.F326B3.98D923.BDA7B6] has joined #code |
| 10:35 | | mode/#code [+o AnnoDomini] by Reiver |
| 11:11 | | Reiv [orthianz@3CF3A5.E1CD01.36D449.95F5A5] has quit [Ping timeout: 121 seconds] |
| 11:17 | | Reiv [orthianz@3CF3A5.E1CD01.36D449.95F5A5] has joined #code |
| 12:07 | | AnnoDomini [annodomini@F67919.F326B3.98D923.BDA7B6] has quit [[NS] Quit: Leaving.] |
| 12:24 | | celticminstrel [celticminst@Nightstar-f8b608eb.cable.rogers.com] has joined #code |
| 12:24 | | celticminstrel [celticminst@Nightstar-f8b608eb.cable.rogers.com] has quit [[NS] Quit: And lo! The computer falls into a deep sleep, to awake again some other day!] |
| 12:41 | | AnnoDomini [annodomini@Nightstar-d990db47.adsl.tpnet.pl] has joined #code |
| 12:41 | | mode/#code [+o AnnoDomini] by Reiver |
| 13:02 | | Rhamphoryncus [rhamph@C06FE3.F5723C.BE3FEB.9D4666] has quit [Client exited] |
| 13:36 | | gnolam [lenin@Nightstar-38637aa0.priv.bahnhof.se] has joined #code |
| 13:36 | < gnolam> | http://www.smbc-comics.com/index.php?db=comics&id=2158 |
| 13:37 | < Tamber> | hehe |
| 13:39 | < gnolam> | Yay, heard back from the GUI guy at last! |
| 13:39 | <@AnnoDomini> | Do you has job? |
| 13:42 | < gnolam> | Possibly. |
| 13:42 | < gnolam> | He has to check with his boss first. |
| 13:43 | < gnolam> | But that's the same boss+1 as the last project, so I think he likes me. :) |
| 14:09 | | Tarinaky [Tarinaky@Nightstar-f349ca6d.plus.com] has joined #code |
| 14:55 | | Attilla [Some.Dude@Nightstar-92c9199f.cable.virginmedia.com] has joined #code |
| 14:56 | | mode/#code [+o Attilla] by Reiver |
| 17:21 | | Kindamoody|out is now known as Kindamoody |
| 17:53 | | Derakon [chriswei@Nightstar-cfae48c3.ca.comcast.net] has joined #code |
| 17:53 | | mode/#code [+o Derakon] by Reiver |
| 17:53 | <@Derakon> | Any recommendations for a Linux recovery CD? |
| 17:53 | <@Derakon> | I tried the Knoppix one and for some reason I can't boot the computer off it. |
| 17:54 | | Kazriko [kaz@Nightstar-5badc7ed.client.bresnan.net] has quit [Client closed the connection] |
| 17:55 | <@AnnoDomini> | Try Damn Small Linux? |
| 17:55 | < Namegduf> | Damn Small Linux has been unmaintained for years |
| 17:55 | <@AnnoDomini> | But it boots from CD. :p |
| 17:56 | < Namegduf> | But it's unmaintained and thus should no longer be suggested for current applications |
| 17:56 | < Namegduf> | It's a dead distribution, no more developers, and has been for a good while |
| 17:56 | <@Derakon> | Hrm, maybe the problem is that I downloaded the "ADRIANE" version, which is apparently blind-friendly? |
| 17:56 | < Namegduf> | Tiny Core Linux is the replacement. |
| 17:57 | | * Derakon goes to try burning another Knoppix CD... |
| 17:59 | <@Derakon> | The layout of ftp://ftp.kernel.org/pub/dist/knoppix/ is just terrible. |
| 18:10 | <@Derakon> | No, that didn't work either. WTF. |
| 18:10 | <@Derakon> | Go to disk utility, select "burn image", choose the Knoppix ISO, computer says burn completed successfully, but the computer won't boot off the damn thing. |
| 18:13 | < Alek> | isn't there usually a checkbox to let it burn as bootable, in the utility? |
| 18:15 | <@ToxicFrog> | Derakon: -any- computer, or just the server in specific? |
| 18:15 | <@ToxicFrog> | What's the failure mode? |
| 18:15 | <@Derakon> | I'm trying to boot up one of our working computers so I can dump its hard drive contents. |
| 18:15 | <@Derakon> | The CD comes before the HDD in the BIOS's boot sequence, and I can see it saying "Boot from CD" in the startup sequence before it gives up and defaults to booting Windows. |
| 18:16 | <@ToxicFrog> | Weird |
| 18:16 | <@ToxicFrog> | Can the same machine boot from other CDs fine? |
| 18:16 | <@Derakon> | This same computer is able to boot from the rescue CD I was using yesterday, but I wasn't able to get network operations working with that one, so I thought I'd try making my own image with a newer OS. |
| 18:16 | <@Derakon> | Haven't tried using my new CD with the computer that has a damage HDD yet, but I doubt it'd work. |
| 18:18 | | * Derakon finds Ubuntu's guide to making a bootable Ubuntu CD, which matches what he was trying to do... |
| 18:18 | <@Derakon> | Guess the next step is to try a different live CD. Any recommendations? |
| 18:18 | <@Derakon> | (Noting the Tiny Core Linux from earlier, though size isn't really a factor here since I have 700MB worth of CD to play with) |
| 18:19 | <@ToxicFrog> | (also, the knoppix image you want is "KNOPPIX_V6.4.4CD-2011-01-30-EN.iso") |
| 18:19 | <@ToxicFrog> | The Ubuntu and SUSE liveCDs are both pretty polished |
| 18:19 | <@Derakon> | Yeah, that's the one I tried second. |
| 18:19 | <@ToxicFrog> | There's an updated version of riplinux, too |
| 18:20 | <@ToxicFrog> | Which IIRC is what you were using earlier |
| 18:20 | <@ToxicFrog> | For smaller stuff, tinycore and puppy are both good choices, although puppy is geared more towards desktop use |
| 18:20 | <@Derakon> | Yeah, RIPLinux is what I was using earlier. |
| 18:22 | <@Derakon> | Guess I'll try Ubuuntu and the updated RIPLinux. |
| 18:23 | <@Derakon> | ...oh. |
| 18:23 | <@Derakon> | DUR. |
| 18:23 | <@Derakon> | Guess what? |
| 18:23 | <@Derakon> | The target computer has a CD drive. |
| 18:23 | <@Derakon> | And I was trying to stick a DVD in it. |
| 18:23 | <@Derakon> | Whups. |
| 18:25 | <@Derakon> | Next step: find a blank CD. Harder than you might think... |
| 18:27 | <@ToxicFrog> | ...I almost asked about that, but then didn't. |
| 18:28 | <@Derakon> | Never assume competency. |
| 18:34 | | Attilla [Some.Dude@Nightstar-92c9199f.cable.virginmedia.com] has quit [Ping timeout: 121 seconds] |
| 18:37 | | RichardBarrell [mycatverbs@Nightstar-228a334c.plus.com] has joined #code |
| 18:53 | <@Derakon> | Okay, the Knoppix CD is pathetically slow. |
| 18:53 | <@Derakon> | Takes something like 10x longer to boot up than the RIPLinux one does. |
| 18:59 | < Namegduf> | It's KDE. |
| 19:00 | < RichardBarrell> | PuppyLinux is the only live media that I've found bearable, and that's because it just sequentially-reads a squashfs into RAM and boots off that. |
| 19:06 | <@Derakon> | Well that was a big waste of time, since Knoppix also did not get the network working automatically. |
| 19:06 | | * Derakon boots back into Windows just to verify that the computer's network is functional at all. |
| 19:06 | < RichardBarrell> | I have a piece of code with 4 coordinate systems in it. :( |
| 19:06 | < RichardBarrell> | Lat/long, tiles, texels and screen pixels. |
| 19:07 | < RichardBarrell> | The only consolation is that tile space is identical to texel space scaled down by 256 and rounded off to ints. |
| 19:12 | < gnolam> | RichardBarrell: at least be glad that you haven't got your lat/long coordinates in arc seconds packed in an obscure format never documented anywhere and only readable by a C program last touched in 1989, whose K&R-iness means that it doesn't even compile anymore. |
| 19:13 | < RichardBarrell> | gnolam: oh, facial tentacles, I'd start praying for the end of the world. |
| 19:13 | < gnolam> | That's only happened to me once though. :P |
| 19:14 | < RichardBarrell> | TBH what I've got here isn't the end of the world. |
| 19:14 | < RichardBarrell> | But it's in javascript. I keep wanting lots of features that JS doesn't give me. |
| 19:15 | | Tarinaky [Tarinaky@Nightstar-f349ca6d.plus.com] has quit [Client closed the connection] |
| 19:15 | < RichardBarrell> | Like I would *love* to have the different numerical variables be tagged with a type that includes their coordinate system, so that I could be certain that I'd never tried to treat a screen pixel as a map texel by mistake or any rubbish like that. |
| 19:19 | <@Derakon> | I forget if Javascript lets you define common math operators on custom objects. |
| 19:23 | < RichardBarrell> | If you define .valueOf() for an object then Javascript will use that to convert it to a number if you attempt to use it in a numerical expression. |
| 19:23 | < RichardBarrell> | Other than that, no, no operator overloading. |
| 19:32 | <@Derakon> | There, finally. New version of RIPLinux was able to get network access through manual configuration, so now I can dump that computer's HDD contents. |
| 19:35 | <@ToxicFrog> | Weirdass. |
| 19:41 | < RichardBarrell> | ToxicFrog: whowhat? |
| 19:41 | <@Derakon> | I suspect he's referring to my having to manually configure the network connection to get it to work. |
| 20:43 | < RichardBarrell> | Yay for Cramer's rule! |
| 20:43 | < RichardBarrell> | (Yes, I know, O(n!), but I'm working specifically on 2x2 matrices.) |
| 20:50 | <@Derakon> | So to copy a disk imiage off of a remote server onto the local computer, I'd do "dd of=/dev/hda < ssh user@host dd if=imageFile" right? |
| 20:51 | <@Derakon> | s/imiage/image/ |
| 20:52 | <@ToxicFrog> | Close. |
| 20:52 | <@ToxicFrog> | < and > are specifically for files. |
| 20:52 | <@Derakon> | Ah. |
| 20:52 | <@ToxicFrog> | ssh user@host dd if=imagefile bs=4M | dd of=/dev/hda bs=4M |
| 20:52 | <@Derakon> | Right. |
| 20:52 | <@Derakon> | Thanks. |
| 20:53 | | celticminstrel [celticminstre@Nightstar-f8b608eb.cable.rogers.com] has joined #code |
| 20:53 | <@Derakon> | I guess I could've enclosed the ssh in backticks or something, but I suspect that could incur some undesirable buffering overhead. |
| 20:53 | <@ToxicFrog> | Um |
| 20:53 | < Namegduf> | Including it in backticks puts the output in as parameters. |
| 20:53 | < Namegduf> | Not on stdin. |
| 20:53 | < Namegduf> | It would, er, be interesting |
| 20:53 | <@Derakon> | Hee. |
| 20:53 | < RichardBarrell> | Heeheee |
| 20:53 | <@Derakon> | Okay, nevermind then~ |
| 20:54 | < RichardBarrell> | echo -n `ssh user@host dd if=/dev/sda1` |
| 20:54 | < RichardBarrell> | There's actually a limit to how long a command-line can be though, sayeth one of the POSIX family standards. It's implementation defined. |
| 20:55 | < Namegduf> | Also it would buffer it in RAM until done. |
| 20:55 | <@Derakon> | Hence the buffering overhead I mentioned. |
| 21:01 | | * Derakon sighs as his boss asks him if he'd "badded" the bad sectors on the hard disk. |
| 21:01 | <@Derakon> | Apparently he thinks that bad sectors are not automatically marked by the hard drive and may be used unless we manually block the computer from doing so. |
| 21:02 | <@Derakon> | You'd think he'd be more mindful of cleanliness in the microscope room and wouldn't talk out of his ass there. |
| 21:02 | <@ToxicFrog> | Er |
| 21:02 | <@ToxicFrog> | That's often true, and is the reason for things like fsck -c |
| 21:03 | <@Derakon> | ...oh, really? |
| 21:03 | <@Derakon> | Well crap. |
| 21:03 | <@Derakon> | Still irrelevant if I'm going to dd over the thing. |
| 21:03 | <@Derakon> | ...right? |
| 21:03 | <@ToxicFrog> | ...not necessarily. |
| 21:03 | <@ToxicFrog> | You dd over the drive. |
| 21:04 | <@ToxicFrog> | Best case: drive remaps bad sectors as you do this and it's fine until more bad sectors start appearing. |
| 21:04 | <@ToxicFrog> | Worst case: drive does not remap bad sectors, filesystem does not realize those sectors are bad (since it's a copy of a filesystem on a good drive) and may attempt to use them. |
| 21:05 | <@Derakon> | So before dding it I should fsck, then. |
| 21:05 | <@Derakon> | Er. |
| 21:05 | <@ToxicFrog> | Um |
| 21:05 | <@Derakon> | I really need to get a better understanding of how hard drives actually work. |
| 21:05 | <@ToxicFrog> | The safest approach is probably: |
| 21:05 | <@Derakon> | And what dd is actually doing. |
| 21:06 | <@ToxicFrog> | - format the drive using the "check for bad blocks" option to your formatting tool, which will scan the drive for bad blocks and store in the filesystem a list of those found so that it won't attempt to use them |
| 21:06 | <@ToxicFrog> | - copy the contents of the good filesystem (rather than the filesystem image itself) onto this drive |
| 21:07 | <@ToxicFrog> | Of course, this fails if, say, the bootloader requires certain files to be located at certain points on the drive |
| 21:07 | <@McMartin> | Isn't that usually handled by having the bootloader get handed a specific partition of its own? |
| 21:11 | < celticminstrel> | Would connectivity problems ever raise a socket.error? |
| 21:12 | < celticminstrel> | Or should I only try to catch socket.timeout? |
| 21:13 | <@ToxicFrog> | celticminstrel: context? |
| 21:13 | <@McMartin> | Nice. I remember reading about these technologies as a grad student. Looks like some prototypes are ready to deploy on full systems. http://code.google.com/p/chronomancer/ |
| 21:14 | < celticminstrel> | My IRC-bot sometimes pings out and fails to reconnect, usually during netsplits. |
| 21:14 | < celticminstrel> | (Not during all netsplits though.) |
| 21:14 | < celticminstrel> | (Well, I'm not sure if it's pinging out. I just know it's disconnecting for some reason.) |
| 21:16 | <@ToxicFrog> | celticminstrel: ...what language? |
| 21:16 | < celticminstrel> | Python, sorry. |
| 21:18 | <@ToxicFrog> | Derakon: if you have specification questions about HDDs/dd/filesystems/badblocks I can try to answer them |
| 21:23 | <@Derakon> | Thanks, TF, but I think this is more "I need to do some reading" than "there is one specific thing I don't understand" right now. |
| 21:25 | | Reiv [orthianz@3CF3A5.E1CD01.36D449.95F5A5] has quit [Ping timeout: 121 seconds] |
| 21:30 | | Reiv [orthianz@3CF3A5.E1CD01.36D449.95F5A5] has joined #code |
| 21:33 | < celticminstrel> | So, if socket.send() or socket.recv() raises a socket.error, should I assume it has disconnected or do I need to look more closely at the exception? |
| 21:36 | < RichardBarrell> | The docs say that socket.error gets sent when one of the syscalls underlying the socket object fails and sets errno. |
| 21:36 | < celticminstrel> | Currently trying to determine how to get at the errno. <_< |
| 21:37 | < RichardBarrell> | try: foo |
| 21:37 | < RichardBarrell> | except socket.error, e: print "Damn! %d." % e.errno |
| 21:38 | < RichardBarrell> | That'll be either due to the universe completely fucking hating you (like ENOMEM - don't even bother considering trying to catch and deal with that), incorrect use of the socket object (ENOTCONN or something - generally indicates a bug in your code, so you should just fix your code rather than trying to handle it)... |
| 21:39 | < celticminstrel> | Is it correct to assume that if recv returns 0 that the connection has been closed? It doesn't feel right. |
| 21:39 | < RichardBarrell> | No, the connection might be half-open. |
| 21:40 | < RichardBarrell> | ...or there are a couple of exceptions that you could conceivably expect |
| 21:40 | < celticminstrel> | ECONNRESET is something I'd want to catch though. |
| 21:40 | <@jerith> | RichardBarrell: Rather use "%s" or "%r" in Python format strings. |
| 21:40 | < RichardBarrell> | like ECONNRESET or EPIPE. |
| 21:41 | < RichardBarrell> | But usually in networking programs, you have only two possible *interesting* states. Either you are able to communicate, or you are not. There is rarely any point trying to discern the exact reason for the latter. |
| 21:41 | <@jerith> | They'll display any data type. "%d" and such are only really required for type-specific modifiers. |
| 21:42 | < RichardBarrell> | jerith: I am aware that %s calls str or unicode as appropriate and %r calls repr, but I still like to have the fact that I believe this thing to be an integer documented in code. |
| 21:43 | < celticminstrel> | So it's reasonable to assume that recv returning 0 means I need to reconnect? |
| 21:43 | <@jerith> | Makes sense. |
| 21:43 | < RichardBarrell> | celticminstrel: lemme check, because I'm not actually sure. |
| 21:43 | <@jerith> | I don't like that it crashes if it isn't an integer, though. :-) |
| 21:43 | < celticminstrel> | And then also catch socket.timeout, plus catch socket.error and check for ECONNRESET... |
| 21:44 | < RichardBarrell> | jerith: I like that it does. If I, the programmer, am mistaken about its type, then there are probably related bugs lurking nearby. A loud, noisy crash alerts me to their presence. :) |
| 21:45 | < RichardBarrell> | A blocking recv(2) returning 0 is only supposed to happen when the other side *deliberately* close(2)'es the socket. |
| 21:45 | < celticminstrel> | Okay, so then it is reasonable to assume it has disconnected. |
| 21:46 | < RichardBarrell> | Yes, and it's reasonable to assume that it did so on purpose. |
| 21:46 | < RichardBarrell> | Otherwise recv(2) would have set errno and returned -1, which will cause Python to throw an exception. |
| 21:47 | < RichardBarrell> | Things get more complicated if you're using non-blocking IO, so don't do that for the moment. ;) |
| 21:48 | < celticminstrel> | I don't think I am. Not really any reason to. |
| 21:49 | <@jerith> | RichardBarrell: Sure, but usually that happens in a log line in an excpetion handler that rarely gets hit and the format string is wrong or the cod'es changed and the type is different from what it was. |
| 21:49 | <@jerith> | But I'm explaining my reasoning, not trying to convince you. :-) |
| 21:50 | < RichardBarrell> | jerith: in the context of logging an exception, I certainly agree with you. |
| 21:53 | <@jerith> | Mostly, I'm just lazy. Using %s or %r means I don't have to thing about the type when I'm displaying it. I can use that tint fragment of brainspace for something else. |
| 21:53 | < RichardBarrell> | I prefer never to allow myself to page out type information. |
| 21:54 | <@jerith> | *tiny |
| 21:54 | < RichardBarrell> | At least, not when I have lots of subtly-different types in bloody adjacent variables. Like numbers representing points in different but isomorphic coordinate systems. |
| 21:55 | <@jerith> | It's not so much paging it out as not having to make a decision about it unnecessarily. |
| 21:55 | < RichardBarrell> | Yes, but making unnecessary decisions about it keep it in the MRU set and stop you from paging it out. :) |
| 21:55 | < RichardBarrell> | *keeps |
| 21:56 | <@jerith> | "Display this thing" is simpler than "display this thing that is an int". :-) |
| 22:07 | | RichardBarrell [mycatverbs@Nightstar-228a334c.plus.com] has quit [Client closed the connection] |
| 22:31 | | Attilla [Some.Dude@Nightstar-92c9199f.cable.virginmedia.com] has joined #code |
| 22:31 | | mode/#code [+o Attilla] by Reiver |
| 23:07 | | AnnoDomini [annodomini@Nightstar-d990db47.adsl.tpnet.pl] has quit [[NS] Quit: Zzz.] |
| 23:27 | | Rhamphoryncus [rhamph@C06FE3.F5723C.BE3FEB.9D4666] has joined #code |
| 23:29 | | shade_of_cpux is now known as cpux |
| 23:42 | <@Derakon> | No going back now. I've started the dd onto the bad hard drive. |
| 23:51 | | Derakon [chriswei@Nightstar-cfae48c3.ca.comcast.net] has quit [[NS] Quit: leaving] |
| 23:52 | | You're now known as TheWatcher[T-2] |
| 23:54 | | You're now known as TheWatcher[zZzZ] |
| --- Log closed Fri Feb 18 00:00:03 2011 |