| --- Log opened Fri Oct 30 00:00:05 2009 |
| 00:17 | | Derakon[AFK] is now known as Derakon |
| 00:23 | | You're now known as TheWatcher[T-2] |
| 00:29 | | Orthia [Orthianz@Nightstar-5aabdcf3.xnet.co.nz] has quit [Client closed the connection] |
| 00:30 | | You're now known as TheWatcher[zZzZ] |
| 00:45 | | Orthia [Orthianz@Nightstar-5aabdcf3.xnet.co.nz] has joined #code |
| 00:48 | < gnolam> | Oh. For. Fuck's. Sake. |
| 00:48 | < gnolam> | Looks like the GNOME issue isn't fixable. |
| 00:49 | <@McMartin> | Hm? |
| 00:49 | < gnolam> | Goddamnit. |
| 00:49 | < gnolam> | Time to ditch Linux altogether. |
| 00:49 | <@McMartin> | Why do you hate freedom so much~ |
| 00:50 | < gnolam> | Because the "freedom" crowd MAKE SHITTY PRODUCTS. |
| 00:50 | < gnolam> | +S |
| 00:50 | <@McMartin> | (More seriously: I missed what the GNOME issue was. Summary?) |
| 00:53 | < gnolam> | Basically: if you press the shutdown/restart/doanythingwithyourONEUSERMACHINE button, it pops up a dialog whose contents are basically |
| 00:53 | < gnolam> | "Are you sure? Are you really sure? Are you really /really/ sure? Oh what the hell, you should have to wait for a full minute before we let you shutdown/restart" |
| 00:53 | < gnolam> | Which is inane, annoying and _fucking stupid_. |
| 00:54 | < gnolam> | And true to the GNOME usability of "We hate the User, so let's disallow him everything we can", they've fucking REMOVED the ability to disable that annoying countdown. |
| 00:55 | < gnolam> | So right now, I'm going "Fuck it, this EEE has been running Linux since I bought it, but this is the last goddamned straw. Where is my old XP key?" |
| 00:56 | < gnolam> | And yes, I've googled it. And in the standard Linux fashion, the answer DOESN'T FUCKING APPLY. |
| 01:01 | <@McMartin> | Awesome. |
| 01:01 | <@McMartin> | Which version of Gnome? |
| 01:04 | <@McMartin> | (Also, Win7 Netbook Edition~) |
| 01:04 | <@McMartin> | (Though that costs $$$ and I haven't seen it in action yet and thus don't know if it's retarded or not) |
| 01:06 | < gnolam> | Whatever version is in Ubuntu Karmic. |
| 01:06 | <@McMartin> | Which I haven't messed with yet. |
| 01:06 | <@McMartin> | Maybe I'll just stick with Jaunty. |
| 01:07 | < gnolam> | There was a reason I stuck to KDE whenever I used Linux, in ye olde days. |
| 01:08 | < gnolam> | But with KDE screwing the pooch so badly, GNOME is pretty much your only officially supported choice nowadays. |
| 01:08 | < Alek> | how'd they screw the pooch? |
| 01:08 | <@McMartin> | Yeah, last time I tried KDE4 it was unusable, and KDE3 was no good for me because it was tied a little too hard to Qt3, which I loathed. |
| 01:08 | <@McMartin> | Qt4 on the other hand is the best cross-platform GUI toolkit I've ever used, so, uh, yay Trolltech, I guess |
| 01:09 | < gnolam> | Yeah, whatever. For me, it's just "Screw all the Linux devs and screw Linux" now. :P |
| 01:09 | < gnolam> | This. Is. IT. |
| 01:09 | <@McMartin> | (Qt4 is actually winning for its Mac support, idly) |
| 01:10 | < gnolam> | Pfft. Mac. |
| 01:10 | <@McMartin> | Yeah, but nobody ever does it right. |
| 01:10 | <@McMartin> | Well, almost. |
| 01:10 | < gnolam> | Apple is among the few companies I feel has done me /personal/ harm. |
| 01:10 | <@McMartin> | The that does is made of spiders and does Windows wrong. =P |
| 01:11 | <@McMartin> | Qt4's the only one I've seen that correctly handles Mac and Windows with a single set of sources. |
| 01:11 | < Alek> | LOL |
| 01:11 | <@McMartin> | And it has a reasonable form designer. Thus: victory. |
| 01:27 | < gnolam> | Oh, there it is. |
| 01:28 | < gnolam> | My old XP key, that is. Not the GNOME preference. Because exposing the latter would actually *GASP* empower the user. |
| 01:28 | < gnolam> | So fuck you GNOME. |
| 01:28 | < gnolam> | Fuck your "We know better than the user, that worthless peon" philosophy. |
| 01:28 | < gnolam> | You just ended over a year of Linux use for me. |
| 01:29 | < gnolam> | Something I'm not incline to try again. |
| 01:29 | < gnolam> | *inclined |
| 01:30 | < gnolam> | I've had a lot of patience with Linux's "quirks". Not again. |
| 02:05 | | Orthia [Orthianz@Nightstar-5aabdcf3.xnet.co.nz] has quit [Client closed the connection] |
| 02:09 | | gnolam [lenin@Nightstar-38637aa0.priv.bahnhof.se] has quit [[NS] Quit: Z?] |
| 02:13 | | Orthia [Orthianz@Nightstar-07df73c0.xnet.co.nz] has joined #code |
| 03:00 | | Attilla [The.Attilla@FBC920.A5C359.15BD32.425F3F] has quit [[NS] Quit: ] |
| 03:06 | | Syloqs-AFH [Syloq@NetworkAdministrator.Nightstar.Net] has quit [Ping timeout: 121 seconds] |
| 03:22 | | Orthia [Orthianz@Nightstar-07df73c0.xnet.co.nz] has quit [Client closed the connection] |
| 03:28 | | Syloqs_AFH [Syloq@NetworkAdministrator.Nightstar.Net] has joined #code |
| 03:29 | | Syloqs_AFH is now known as Syloqs-AFH |
| 03:30 | | Orthia [Orthianz@Nightstar-ed39d757.xnet.co.nz] has joined #code |
| 03:34 | <@Kazriko> | ... I use UNR and I haven't noticed a problem like what he's describing... |
| 03:34 | <@Kazriko> | and it's gnome based... |
| 03:39 | | Syloqs-AFH [Syloq@NetworkAdministrator.Nightstar.Net] has quit [Ping timeout: 121 seconds] |
| 03:43 | | Syloqs_AFH [Syloq@NetworkAdministrator.Nightstar.Net] has joined #code |
| 03:44 | | Syloqs_AFH is now known as Syloqs-AFH |
| 04:02 | | Syloqs-AFH [Syloq@NetworkAdministrator.Nightstar.Net] has quit [Connection reset by peer] |
| 04:32 | <@ToxicFrog> | Kaz: I have been using GNOME for years. I have never noticed this either. |
| 04:33 | <@ToxicFrog> | Or, well |
| 04:33 | <@McMartin> | It's on stock Ubuntu. When you push "Poweroff" it pops a countdown dialog box. |
| 04:33 | <@ToxicFrog> | On Ubuntu - and OpenSUSE, for that matter - when I try it, it pops up a window with options "log out", "suspend", "hibernate", and "poweroff" |
| 04:34 | < Namegduf> | Wait |
| 04:34 | < Namegduf> | You're using Ubuntu |
| 04:34 | <@ToxicFrog> | And a 60-second timer at the bottom so if the press the button and then go make tea, it does something sensible (ie, turns off) |
| 04:34 | < Namegduf> | And you're surprised when it tries to take idiot proofing to, well, insane levels? |
| 04:34 | <@ToxicFrog> | I'm not sure whether to read gnolam's complaints as "how dare it ask me what I want to do, it should read my mind" |
| 04:34 | <@McMartin> | TF: Yes, the 60-second timer is what he's objecting to; he wants immediate poweroff and can't set a config option to do this. |
| 04:34 | <@ToxicFrog> | Or as "how dare it ask me what I want to do, it should shut down instantly" |
| 04:35 | <@McMartin> | The latter. |
| 04:35 | <@ToxicFrog> | Or as "how dare it have a timer, it should damn well sit there until I'm ready to decide" |
| 04:35 | <@McMartin> | The middle. |
| 04:35 | <@McMartin> | I suspect the objection is that if you, instead of popping the dialog box you name, actually select "poweroff" directly, it still pops the timer. |
| 04:35 | <@ToxicFrog> | Hmm. System->Preferences->Power Management->General: When the power button is pressed: Shut Down |
| 04:35 | <@ToxicFrog> | Aah. |
| 04:36 | <@McMartin> | And that the option you named was apparently dropped in 9.10 |
| 04:36 | <@ToxicFrog> | I'm using opensuse here, actually~ |
| 04:36 | | * McMartin recalls something similar in 9.04, anyway~ |
| 04:37 | <@McMartin> | Ever since I single-booted Spiff and Zinglon I haven't really used Linux as anything but an SSH target. |
| 04:37 | <@ToxicFrog> | ...hmm. I can't find my multiOS liveUSB :( |
| 04:39 | | * Namegduf uses Linux as his main OS, but also doesn't use Ubuntu |
| 04:41 | <@McMartin> | I'd dispute your line about "idiot-proofing"; it's extremely unfriendly even a little bit below the surface because it's fundamentally Debian-based, and Debian hates you and wants you to die. |
| 04:41 | < Namegduf> | No it doesn't. :P |
| 04:41 | <@ToxicFrog> | Yes. It does. ;.; |
| 04:41 | <@ToxicFrog> | Or at least it did when last I tried it. |
| 04:41 | < Namegduf> | Was that pre or post 2000? |
| 04:41 | <@ToxicFrog> | 2007, I think. |
| 04:42 | <@ToxicFrog> | Might have been '08. |
| 04:42 | <@McMartin> | The usual hilarious adventure involves trying to install subversion. |
| 04:42 | | * ToxicFrog also uses Linux as his primary OS, but uses OpenSUSE, mostly |
| 04:42 | <@ToxicFrog> | McMartin: Python, for me |
| 04:42 | < Namegduf> | McMartin: apt-get install subversion |
| 04:42 | <@McMartin> | WRONG |
| 04:42 | < Namegduf> | No, not wrong |
| 04:42 | <@McMartin> | if you do that, you will not in fact be able to svn co anything |
| 04:42 | <@McMartin> | You must apt-get install svn-client |
| 04:42 | <@ToxicFrog> | # apt-get install python |
| 04:42 | <@ToxicFrog> | Ok! |
| 04:42 | <@ToxicFrog> | $ program that depends on python |
| 04:42 | <@ToxicFrog> | BLARGH YOU DON'T HAVE PYTHON |
| 04:42 | < Namegduf> | McMartin: "Yes, you can" |
| 04:42 | <@McMartin> | "subversion" is the server. |
| 04:43 | | * Namegduf checks on his Debian server, on which svn's client is used |
| 04:43 | <@ToxicFrog> | <half an hour of research reveals that 'python' is in fact python 2.3, and I want 'python25', because god forbid the package manager keep track of versions) |
| 04:43 | <@McMartin> | Namegduf: In neither TF's, my, nor one other person in #code we had to talk through this, was this the case. |
| 04:43 | <@ToxicFrog> | # apt-get install python25 |
| 04:43 | <@ToxicFrog> | BLARGH THOUSANDS OF ERRORS EVERYWHERE |
| 04:44 | <@ToxicFrog> | ...fast forward a few hours, in which I've removed python, still can't install python25, and am hip-deep in manually resolving dependencies for a source build of python 2.5 |
| 04:44 | < Namegduf> | "This package includes the Subversion client (svn), tools to create a Subversion repository (svnadmin) and to make a repository available over a network (svnserve)" |
| 04:44 | <@ToxicFrog> | At which point I realized it would be faster to install a different distro |
| 04:44 | < Namegduf> | ~ aptitude on "subversion" |
| 04:44 | <@ToxicFrog> | So I did, and I was happy. |
| 04:44 | <@ToxicFrog> | Not very happy, since it was Gentoo, but happier than when I was using Debian~ |
| 04:44 | <@McMartin> | Namedguf: As it happens, the install never actually worked for any of us. |
| 04:45 | < Namegduf> | Which install? Debian's? |
| 04:45 | <@McMartin> | Yeah. |
| 04:45 | <@McMartin> | In fact |
| 04:45 | <@McMartin> | ISTR that being part of TF's saga of trying to build python 2.5 |
| 04:45 | | * Namegduf had the same experience with the newer Ubuntu ISOs, but ONLY the newer ones, because they always had to bring up X on his laptop and failed at it. |
| 04:45 | < Namegduf> | Debian could actually bring up X, *and* included the capacity not to. |
| 04:45 | <@McMartin> | In my case, this was the Debian repos underlying Ubuntu. |
| 04:46 | <@ToxicFrog> | ...either Debian or Gentoo had an install-via-ssh mode that actually worked great for me |
| 04:46 | <@ToxicFrog> | But I can't remember which one it was |
| 04:46 | < Namegduf> | Ubuntu's repos are largely Debian's, but with more reliability issues than Debian Testing |
| 04:46 | <@McMartin> | Gentoo had one, but your definition of "worked great" has to be scaled to gentoo. |
| 04:46 | < Namegduf> | Or removed. |
| 04:46 | | * Namegduf used to use Gentoo. A long time ago. |
| 04:46 | <@McMartin> | Likewise. |
| 04:46 | < Namegduf> | Back when the repositories were competently maintained. |
| 04:46 | <@McMartin> | I stopped when I realized that it was incapable of handling upgrades. |
| 04:46 | < Namegduf> | Oh, and that, too. |
| 04:46 | < Namegduf> | I dropped it moved to GoboLinux, larger moved to Debian Testing. |
| 04:47 | <@ToxicFrog> | I used it for a few days, and then decided "fuck $USE" |
| 04:47 | <@McMartin> | This was in... oh, 2005, I believe. |
| 04:47 | <@McMartin> | Maybe 2004 |
| 04:47 | < Namegduf> | Installed gdm and XFCE, and all was well. |
| 04:47 | <@McMartin> | I switched over to FC3, which was at the time current. |
| 04:47 | <@ToxicFrog> | Namegduf: re ubuntu X installers: isn't there an 'ubuntu-alternative-install' CD or something with a similar name that includes various S3 and tty modes? |
| 04:47 | < Namegduf> | And... python2.5, why are you complaining that a development-line Python was hard to install? |
| 04:48 | <@McMartin> | Um |
| 04:48 | < Namegduf> | Wasn't 2.5 a dev branch? |
| 04:48 | <@McMartin> | No |
| 04:48 | <@McMartin> | Not even remotely. |
| 04:48 | < Namegduf> | Huh. |
| 04:48 | < Namegduf> | Okay. |
| 04:48 | <@ToxicFrog> | Namegduf: because I had a list of programs as long as my arm that needed it, and it had in fact been out for quite some time and had been production quality for most of that? |
| 04:48 | <@ToxicFrog> | And was in fact standard on all of my non-Debian systems? |
| 04:48 | <@McMartin> | 2.5 is standard as shovelware on Dell XP machines. |
| 04:48 | < Namegduf> | What, now? |
| 04:49 | <@ToxicFrog> | (well, except for Durandal, but Durandal runs XP) |
| 04:49 | <@McMartin> | (They use it to manage their OEM stuff) |
| 04:49 | | * Namegduf assumed it was dev because everyone seems to be using either 2.4 or 2.6 |
| 04:49 | < Namegduf> | Only references to 2.5 I've found are "This changed in 2.5" |
| 04:49 | <@McMartin> | 2.4 is painfully obsolete, 2.6 is 2.5 with some nods to supporting 3.0 stuff. |
| 04:49 | | * Namegduf has no issues with 2.4 |
| 04:50 | <@McMartin> | Lucky you. |
| 04:50 | <@ToxicFrog> | I suspect everyone is using 2.4 because some stuff written for 2.4 doesn't work in 2.5+, or 2.6 because it supercedes 2.5. |
| 04:50 | <@McMartin> | 2.5 dramatically increases the available libraries and adds some __future__ imports that unbreak a lot of things. |
| 04:50 | <@ToxicFrog> | Which obviously wasn't the case when 2.6 hadn't been released yet. |
| 04:50 | <@McMartin> | Quite. |
| 04:51 | <@ToxicFrog> | Oh yes, this reminds me, I need to make some changes to my LuaPilot bindings |
| 04:51 | < Namegduf> | 2.4 is primarily, the oldest one people actually seem to care about supporting nowadays. |
| 04:51 | <@ToxicFrog> | (which is now officially sanctioned by the university pending these changes!) |
| 04:51 | < Namegduf> | At least, that's the impression I get. |
| 04:51 | < Namegduf> | So it's nice to develop with. :P |
| 04:51 | <@McMartin> | See, distros that aren't Debian have a single "python" package. |
| 04:51 | <@McMartin> | Which updates when there are updates. |
| 04:51 | < Namegduf> | That's far too overly vague. |
| 04:52 | <@McMartin> | This is exactly the mindset I despise and seek to destroy~ |
| 04:52 | < Namegduf> | There's distros out there which don't have the ability to install older branches and get updates on them properly at all. |
| 04:52 | < Namegduf> | Which leaves everyone who has existing software they want to Just Work and get updates stably up shit creek. |
| 04:52 | <@McMartin> | Fedora, incidentally, solves this by letting you select a version when you check something out as an argument. |
| 04:52 | <@ToxicFrog> | So does OpenSUSE. |
| 04:52 | <@McMartin> | In Debian, if you want the latest version of something, you must know. |
| 04:52 | < Namegduf> | And Debian puts it in the package name. |
| 04:53 | < Namegduf> | No, in Debian, if you want the *non-default* version, you must know. |
| 04:53 | < Namegduf> | That may or may not be the most recent version. |
| 04:53 | <@McMartin> | And its defaults are universally horribly wrong. |
| 04:53 | <@ToxicFrog> | Yes, which means that for one logical package, you have multiple actual packages, one for each version, all of which conflict with each other |
| 04:53 | <@McMartin> | It also means that when a new version comes out you don't receive it. |
| 04:53 | < Namegduf> | Right. And that's logically equivalent to one package, multiple versions. |
| 04:53 | < Namegduf> | And? |
| 04:53 | <@McMartin> | ^-- |
| 04:53 | <@ToxicFrog> | And the one that's just 'packagename' rather than 'packagename1234' or 'packagename-1234' or 'packagename-1.2.3.4' is never the version you need, because it's four years out of date. |
| 04:53 | < Namegduf> | You don't WANT to receive it by default. |
| 04:54 | <@McMartin> | YES YOU DO, this is what keeping up-to-date means |
| 04:54 | <@ToxicFrog> | YES, YOU DO |
| 04:54 | < Namegduf> | If I am using something for production shit, I do *not* want it to jump to new major branches for no reason. |
| 04:54 | < Namegduf> | I want it to get bugfixes on the *existing*, *still supported and maintained* branches. |
| 04:54 | < Namegduf> | ESPECIALLY when that has the possibility to break my production environment. |
| 04:54 | <@ToxicFrog> | Namegduf: surprisingly, most people are not running production servers that require a signoff by IT before anything gets updated |
| 04:55 | < Namegduf> | ToxicFrog: No, I'm not, either. |
| 04:55 | <@McMartin> | And the ones that do can select the version they want to go to. |
| 04:55 | <@McMartin> | Unless, of course, the package maintainer decided that *this* bugfix was significant enough to deserve a new package name, and you want that. |
| 04:55 | < Namegduf> | ToxicFrog: I don't have a "Whee, let's break everything and fix it up later" attitude, either. |
| 04:55 | <@McMartin> | And you'll never know unless you regularly search the entire repository. |
| 04:55 | <@ToxicFrog> | Namegduf: and yet you use Debian~ |
| 04:56 | < Namegduf> | ToxicFrog: Right. Debian doesn't do that. That's what you were just complaining about. |
| 04:56 | < Namegduf> | Well, they *do*, but in sid |
| 04:56 | <@ToxicFrog> | Er |
| 04:56 | <@ToxicFrog> | Ok, if you define "actually being able to install software released this year" as "breaking things" then yes, Debian does not do that |
| 04:57 | <@McMartin> | See, if Program B needs to be patched to work with an update of library A, either B is marked to require a specific version of library A (less common) or B is updated in the repos at exactly the time A is (more common). |
| 04:57 | <@McMartin> | The latter is notable for meaning that you type "yum update" and you have an updated system that continues to work. |
| 04:57 | < Namegduf> | Right. |
| 04:57 | < Namegduf> | Debian does the same. |
| 04:57 | < Namegduf> | Unless you're in Sid. |
| 04:57 | < Namegduf> | In fact, one of the major defining things of Testing is that dependency issues get ironed out before stuff comes out of Sid. |
| 04:58 | < Namegduf> | There's "exceptions", but not on Stable. |
| 04:58 | <@McMartin> | I'm unfamiliar with its ability to handle updates because I can almost never get stuff installed in the first place. |
| 04:58 | < Namegduf> | If you can't use aptitude, you have issues |
| 04:58 | <@McMartin> | Stable is five years or more out of date; this is unacceptable for a dev environment. |
| 04:58 | | * SmithKurosaki prods TF |
| 04:58 | < Namegduf> | McMartin: That's just ridiculous crap |
| 04:58 | <@McMartin> | aptitude has never worked remotely as well as yum. |
| 04:58 | <@ToxicFrog> | So, having three different packages for the same program, all with different names, is not "a dependency issue"? |
| 04:58 | < Namegduf> | Stable only exists for 18 months |
| 04:59 | <@McMartin> | The last time I looked at Stable it was using Gtk 2.0.0. |
| 04:59 | <@McMartin> | Gtk was, at the time, at least 14 releases past that. |
| 04:59 | <@ToxicFrog> | Because it damn well seemed to be when I tried to install stuff that depended on having a modern version of Python. |
| 04:59 | < Namegduf> | 2.12.12 |
| 04:59 | <@McMartin> | Testing was at the time at 2.8 something. |
| 04:59 | <@McMartin> | However, I got reports that I didn't run on Debian Stable becuase it relied on calls introduced in 2.1.x |
| 05:00 | < Namegduf> | It's at most a year and a half "out of date", where "out of date" means "bugfixes and security updates only, no major new features" |
| 05:00 | <@McMartin> | What was, at the time, seven full releases back, not counting point releases. |
| 05:00 | <@McMartin> | Or minor new features, or API calls that have been standard for over a year and thus are used in many programs that you apparently do not deserve to use |
| 05:01 | < Namegduf> | I've never observed any real, useful programs jumping to use new APIs that quickly. |
| 05:01 | < Namegduf> | Minor new features Do Not Belong In Stable. |
| 05:02 | <@McMartin> | Gtk 2.0 was very, very, very, very, very bad. |
| 05:02 | < Namegduf> | It's for running things you want to quietly update without babysitting them constantly, and to Do Their Damn Job. |
| 05:02 | < Namegduf> | It's not for screwing around |
| 05:02 | <@McMartin> | It didn't even become slightly usable until 2.4 |
| 05:02 | < Namegduf> | For that, go run Testing/Sid |
| 05:02 | <@McMartin> | Or Fedora, or SUSE, or Ubuntu, none of which had this problem. |
| 05:02 | < Namegduf> | Ubuntu is less stable than Testing |
| 05:02 | <@McMartin> | Only when you go under the hood. |
| 05:03 | <@ToxicFrog> | Testing and Sid still have the "different packages for different versions" problem |
| 05:03 | < Namegduf> | That's not a "problem". |
| 05:03 | <@McMartin> | And again, the stablest program has zero lines, and this is exactly the problem I had with Debian. |
| 05:03 | < Namegduf> | It does not impact on the system meeting its requirements and doing its job. |
| 05:03 | <@ToxicFrog> | Yes. It does. |
| 05:03 | <@McMartin> | It impacts the ability of the system to be constructed in the first place. |
| 05:03 | <@ToxicFrog> | Because my requirements are "I can install and run software" |
| 05:03 | < Namegduf> | I've never had a problem with doing that. |
| 05:03 | <@McMartin> | Then your requirements are significantly below mine. |
| 05:03 | < Namegduf> | No, they're not. |
| 05:03 | <@ToxicFrog> | And it does not meet this requirement as soon as the software I need to install and run is a non-default version of something. |
| 05:04 | <@McMartin> | Yes, yes they are. |
| 05:04 | <@ToxicFrog> | It either breaks when I try to install it; or it installs it, and in so doing breaks something else. |
| 05:04 | < Namegduf> | Packages with version numbers in the name are equivalent, precisely, to different versions of the same package which cannot be installed alongside each other. |
| 05:04 | <@McMartin> | As evidenced by the fact that I needed to do development work on a specific program that used SDL - not remotely new, even then - and could not install the requisite libraries and tools without going to "experimental". |
| 05:04 | < Namegduf> | They are quite literally identical in functionality. |
| 05:04 | < Namegduf> | Any operation with one can be mapped onto the other. |
| 05:04 | <@ToxicFrog> | Namegduf: no. They are not. |
| 05:05 | < Namegduf> | I don't think Experimental even is used anymore |
| 05:05 | <@McMartin> | Not without adding a chain of dependencies between packages to indicate "the next version is this one", which is lacking. |
| 05:05 | <@ToxicFrog> | In openSUSE, or Fedora, or Lunar, or Ubuntu, I can say "update python" and it will do so. |
| 05:05 | <@ToxicFrog> | In Debian, I can say "update python" and it will say "it's already 2.4, which is the latest version" |
| 05:05 | < Namegduf> | Between major branches? |
| 05:05 | < Namegduf> | So if you manually request 2.3 |
| 05:05 | < Namegduf> | And update it, will it switch to 2.5? |
| 05:05 | <@ToxicFrog> | Because what I actually need to do is remove python and install python25 |
| 05:05 | < Namegduf> | Because that's not equivalent, no, it's less functional. |
| 05:05 | < Namegduf> | By far.. |
| 05:06 | <@ToxicFrog> | Namegduf: it will say "this will update python 2.3 to python 2.5, proceed y/n?" |
| 05:06 | < Namegduf> | Because it means the ability to install older branches and still update is broken. |
| 05:06 | <@ToxicFrog> | Or, if I really need python 2.3, I say "pin python" |
| 05:06 | < Namegduf> | And then it doesn't get updates at all? |
| 05:06 | < Namegduf> | That's really unacceptable |
| 05:06 | < Namegduf> | I want Python 2.4.x |
| 05:06 | <@ToxicFrog> | THAT'S WHAT DEBIAN ALREADY DOES |
| 05:06 | < Namegduf> | No, it isn't. |
| 05:06 | <@ToxicFrog> | Except it's worse at it! |
| 05:06 | < Namegduf> | Because, TF, you may have missed it |
| 05:06 | <@McMartin> | 2.3 hasn't been updated in years. |
| 05:06 | <@ToxicFrog> | Because I can't just unpin, or ask it specifically to updayte |
| 05:06 | < Namegduf> | But Python has more than two numbers in its version string |
| 05:07 | <@ToxicFrog> | Instead I need to search the entire repo to find out what the real latest version |
| 05:07 | <@ToxicFrog> | is |
| 05:07 | < Namegduf> | Yeah /python is really hard |
| 05:07 | <@ToxicFrog> | And remove one package (which may cause the dependency resolver to entirely lose its shit) and install the other |
| 05:07 | < SmithKurosaki> | And remove one package (which may cause the dependency resolver to entirely lose its shit) and install the other |
| 05:07 | < Namegduf> | That's pretty invalid. |
| 05:07 | < SmithKurosaki> | That has happened to me before |
| 05:07 | < Namegduf> | I mean, any distro can have bugs like that happen. |
| 05:07 | < Namegduf> | And in my experience, Debian's a lot less prone to them. |
| 05:08 | < SmithKurosaki> | nini |
| 05:08 | <@McMartin> | Namedguf |
| 05:08 | < Namegduf> | And, see, when I *want to manually alter the state*, having to *manually search for a new branch version* is acceptable |
| 05:08 | <@McMartin> | I just typed /python into Aptitude |
| 05:08 | <@McMartin> | I then had to hit n 94 times to actually reach "python2.6" |
| 05:09 | <@McMartin> | That is the easiest thing ever; I stand corrected. |
| 05:09 | < Namegduf> | When I want to perform routine updates, having to *manually check for new versions*, or *manually tell it not to update automatically between major branches* |
| 05:09 | < Namegduf> | Is not acceptable |
| 05:09 | < Namegduf> | My Python 2.4.x still needs to get its bugfixes. I just don't want to find out I'm suddenly running 2.6 and my Python-based web stuff that's rather important has all broken. By default. |
| 05:09 | <@McMartin> | Having to do it by hand for minor branches is not acceptable. |
| 05:10 | < Namegduf> | They aren't minor branches. |
| 05:10 | < Namegduf> | They're not often released, they're the main releases. |
| 05:10 | <@ToxicFrog> | You seem to define "routine update" as "I only want security or bugfix updates to anything; feature updates I expect to do by hand on a per-package basis" |
| 05:10 | < Namegduf> | Bingo! |
| 05:10 | <@ToxicFrog> | You are in the minority on this. |
| 05:10 | <@McMartin> | This is, in fact, madness on stilts. |
| 05:10 | < Namegduf> | No, I'm just not treating my servers like a toy. |
| 05:11 | <@ToxicFrog> | Hey, I have five linux machines, exactly one of them is a server and it's not public-facing. |
| 05:11 | < Namegduf> | If I want shiny new things, I'm quite fine with asking for them. In general, however, their job is to sit down, shut up, and serve crap |
| 05:11 | < Namegduf> | And take as little of my time in maintenance overhead as possible, so I can do better thigns. |
| 05:11 | <@McMartin> | See, I never got that far. |
| 05:11 | < Namegduf> | I run Debian Testing on my laptop, because there I *do* want shiny newness |
| 05:12 | <@McMartin> | It wasn't until I was forced to handle it for someone else that I got far enough to find the 12 packages or so you that you need to have what any other distro would call "install gcc" |
| 05:12 | < Namegduf> | And am willing to deal with suddenly running new versions of X and recompiling my downloaded-so-I-have-the-literal-latest nVidia drivers for them |
| 05:12 | <@McMartin> | Suddenly? |
| 05:13 | <@McMartin> | I admit I use synaptic to handle upgrades with. Does aptitude not tell you what it's going to upgrade? |
| 05:13 | < Namegduf> | Er, it does. |
| 05:13 | < Namegduf> | And, 12 packages for compiling is a lie. |
| 05:13 | < Namegduf> | Sorry, but I compile on ALL my Debian servers |
| 05:13 | < Namegduf> | It *is* just installing gcc |
| 05:13 | <@McMartin> | You also need -doc. |
| 05:13 | <@McMartin> | At minimum. |
| 05:14 | <@McMartin> | And then there's a bunch of other ones it recommends. |
| 05:14 | <@ToxicFrog> | To be fair -doc is a seperate package on OpenSUSE and I think Fedora, too. |
| 05:14 | < Namegduf> | You don't need to install recommendations. They're cool suggestions, that's all. |
| 05:14 | < Namegduf> | Largely, I don't. |
| 05:14 | <@ToxicFrog> | And gcc/g77/g++/etc are generally seperate packages, too. |
| 05:14 | <@McMartin> | Fedora tends to fold -doc into the -devel package. |
| 05:14 | < Namegduf> | Yeah, they are, which is a Good Thing, because they're kinda big |
| 05:14 | <@McMartin> | You get the man pages with the headers. |
| 05:14 | <@ToxicFrog> | That's not unique to Debian and I don't really consider it a problem, either. |
| 05:14 | < SmithKurosaki> | manpages-dev |
| 05:14 | < SmithKurosaki> | Always nice to have |
| 05:14 | < Namegduf> | You might, hmm |
| 05:14 | <@ToxicFrog> | Although I do prefer the fedora approach to man pages. |
| 05:14 | < Namegduf> | "make" |
| 05:15 | < Namegduf> | "autoconf" if you're rolling your own build systems... |
| 05:15 | <@ToxicFrog> | If you mean "I need twelve packages, which are not autoresolved, to use 'gcc'", that is a problem, but I wouldn't know because I never got the goddamn machine to the point of installing gcc on it |
| 05:15 | < Namegduf> | How? |
| 05:15 | < Namegduf> | Did it not get a network connection? |
| 05:15 | <@McMartin> | TF: No, I meant "to get what I consider an appropriate dev environment where all my support commands work" |
| 05:16 | <@McMartin> | "... not counting stuff like text editors or test harnesses" |
| 05:16 | < Namegduf> | Is there a distro other than Gentoo (which needs it) which ships with all that? |
| 05:16 | <@ToxicFrog> | Namegduf: I spent three hours trying to get python 2.5 installed so that critical applications I required would run at all, and then gave up and installed Gentoo, which was horrible, but still more enjoyable and more functional than Debian |
| 05:17 | <@ToxicFrog> | After a few days of that and once the immediate crisis was handled, I wiped Gentoo, too, and installed Lunar, which is Gentoo without the horrible. |
| 05:17 | <@ToxicFrog> | (Oculus is still running it) |
| 05:17 | < SmithKurosaki> | I assume Debian and not just a debian derivative |
| 05:17 | | * Namegduf is not fond of any Gentoo developers, ever |
| 05:17 | <@ToxicFrog> | Yes, Debian. |
| 05:18 | <@ToxicFrog> | Namegduf: everyone involved with USE flags owes me a personal apology |
| 05:18 | < Namegduf> | The impression I have gotten is that they are with little exception drama whores, lying drama whores, or they quit eons ago. |
| 05:18 | <@McMartin> | Namegduf: Fedora and Ubuntu both lay down all you need if you check "install basic dev tools" during installation. |
| 05:18 | < SmithKurosaki> | nini |
| 05:18 | <@McMartin> | Gives you at minimum the autotools, gcc, g++, and the related libs, libs-devel, and headers. |
| 05:19 | < Namegduf> | Does that even count? |
| 05:19 | <@McMartin> | I'm inclined to count it. |
| 05:19 | < Namegduf> | I've found those things to have all the precision and usefulness of a nuclear bomb, on all distributions. |
| 05:19 | < Namegduf> | Er, no, that only covers the precision part. |
| 05:19 | <@McMartin> | I've had excellent track records with it on both. |
| 05:20 | <@McMartin> | Of course, Fedora does this right all around, so if I'm trying to compile something and need a library, I go "yum install (libraryname)-devel" and then proceed, now with appropriate references and headers, as well as a runtime that stays synched. |
| 05:20 | < Namegduf> | Yeah, Debian does that too. |
| 05:20 | <@McMartin> | No, I need a version number. |
| 05:20 | < Namegduf> | It, horrifying, dares to call it -dev instead of -devel |
| 05:20 | <@McMartin> | And I need -dev and -doc. |
| 05:21 | < Namegduf> | Maybe -doc, I've never bothered. |
| 05:21 | < Namegduf> | I don't rely on manpages for my API documentation. |
| 05:21 | < Namegduf> | And you don't need a version number unless it's incompatible between versions. |
| 05:21 | < Namegduf> | In which case, you probably need the same version your software is designed to compile against. |
| 05:21 | < Namegduf> | So it's not really a bad thing, IMO. |
| 05:22 | <@McMartin> | I have never, ever, found the default version to be modern enough for any library I've ever tried to compile against, ever. |
| 05:22 | <@McMartin> | Not even for SDL. |
| 05:22 | | * Namegduf has never had issues. Ever. |
| 05:23 | <@McMartin> | You will note my continuing astonishment with respect to this. |
| 05:23 | < Namegduf> | I don't try to use the newest, shiniest features of everything in my code, and I've found exceedingly little third-party code that moves ahead in API versions that quickly. |
| 05:23 | <@McMartin> | Which is also why I suspect your requirements were either more in line with their goals, or were very low. |
| 05:23 | < Namegduf> | (And yes, "that quickly" means last year or two) |
| 05:24 | <@McMartin> | Yeah, GUI libraries evolve a hell of a lot faster than that, largely because none of them are acceptable. |
| 05:24 | < Namegduf> | I run Debian Testing on my laptop, so no comments there. |
| 05:25 | <@McMartin> | You seem to have this preconception that anything that involves direct interaction with an application is using the machine as a toy. This is not true for all use cases. |
| 05:25 | < Namegduf> | Well... there's very few practical cases where you *absolutely must have this new API/app feature for production use now*. |
| 05:26 | < Namegduf> | I have a few myself, but "now" in those cases precludes waiting for it to get into any repo. |
| 05:26 | < Namegduf> | And I compile myself. |
| 05:26 | < Namegduf> | (Also, I have a bunch of patches and my own git repos for most of it) |
| 05:27 | <@ToxicFrog> | For you, "production use" appears to be "I need a stable, world-facing web server that will never do anything without my say-so" |
| 05:27 | < Namegduf> | Bingo, aside get security updates and such and keep working. |
| 05:27 | <@ToxicFrog> | For the rest of us, it's "hey, I want a working desktop with modern software, and possibly the ability to compile my programs" |
| 05:27 | < Namegduf> | Production stuff is generally already working, and already working software rarely needs new features in its libraries. |
| 05:27 | <@McMartin> | My default "production use" is "I am deploying an application to several dozen only moderately tech-savvy office workers." |
| 05:27 | <@ToxicFrog> | This may be where a lot of the differences in experience comes from. |
| 05:28 | < Namegduf> | The "lack of shiny new things" might be too painful for a non-corporate desktop, yeah. |
| 05:28 | <@McMartin> | Even corporate ones. |
| 05:28 | <@ToxicFrog> | I use linux as either (a) a development laptop, (b) a development laptop, or (c) a personal everythingserver. |
| 05:28 | <@ToxicFrog> | Debian fails hard at all of these. |
| 05:28 | <@ToxicFrog> | Er. |
| 05:28 | <@ToxicFrog> | s/laptop/desktop/ |
| 05:28 | < Namegduf> | *Stable* does, yeah. |
| 05:28 | <@McMartin> | You listed "development laptop" twice |
| 05:28 | < Namegduf> | I have... hmm. |
| 05:29 | < Namegduf> | Four Debian servers in production use to varying degrees. |
| 05:29 | < Namegduf> | All are net-facing |
| 05:29 | <@McMartin> | I've actually gotten to the point where I use Linux solely for text services and data archiving because Windows or OS X meet all my other goals far better, and "tri-platform compatibility" being a primary dev goal means that I have ready access to all three. |
| 05:29 | <@ToxicFrog> | Namegduf: so does unstable due to aforementioned issues, and I preemptively respond to your "but having multiple packages for each thing is a feature" with "only in the aforementioned deployed-and-untouched server use case, which you may have noticed is not how I use any of my computers, ever" |
| 05:30 | < Namegduf> | Right. |
| 05:30 | < Namegduf> | Fair enough, I suppose. |
| 05:30 | < Namegduf> | I know Testing moves between default versions from time to time. |
| 05:30 | < Namegduf> | Not with Python, but it's actually managed to break some of my stuff doing that. |
| 05:30 | <@ToxicFrog> | Debian may be good for that; it may even be ideal for that; but it's not how I, or most of the people I know, use their computers. |
| 05:30 | <@ToxicFrog> | And attempting to use Debian the way we actually use our computers results in pain and horror. |
| 05:30 | < Namegduf> | Yeah; it's not really how you use "your" computer. |
| 05:31 | <@McMartin> | It's also not how anyone I've ever developed for uses theirs. |
| 05:31 | < Namegduf> | (That is to say, any personal machines) |
| 05:31 | <@ToxicFrog> | Whereas OpenSUSE, Fedora, and Ubuntu tend to be kittens all the way down. |
| 05:31 | <@McMartin> | No, Ubuntu is three layers of kittens and then spiders. |
| 05:31 | < Namegduf> | I have dealt with an Ubuntu server before |
| 05:31 | <@ToxicFrog> | (well, until you hit the audio subsystem~) |
| 05:31 | < Namegduf> | It made me want to die |
| 05:31 | < Namegduf> | It was like Debian, but actually *missing* dev packages I needed. |
| 05:31 | <@ToxicFrog> | I wouldn't care to use an Ubuntu server. |
| 05:31 | < Namegduf> | I think it lacks "htop" |
| 05:31 | <@McMartin> | Ubuntu is probably the best Linux desktop OS, but that's what it optimizes for. |
| 05:32 | <@ToxicFrog> | I've used Fedora and OpenSUSE as both desktop and server machines without problems. |
| 05:32 | < Namegduf> | What the hell is more friendly about lacking htop? |
| 05:32 | < Namegduf> | Yeah. |
| 05:32 | < Namegduf> | Ubuntu is very much not good for what Debian is. |
| 05:32 | <@McMartin> | Fedora and SUSE (but I couldn't get past the KDE4 interface) try to be all things to all people and are mostly successful at it. |
| 05:32 | <@ToxicFrog> | Most of the nonprogrammers I know using Linux are using Ubuntu, and seem happy with it. |
| 05:32 | <@ToxicFrog> | (SUSE has a GNOME version, which is IIRC the default) |
| 05:32 | < Namegduf> | I, personally, prefer Debian Testing for desktop use, too, but that's because I view GNOME as a sin against nature and know how to install gdm and XFCE manually |
| 05:32 | <@McMartin> | Fedora in particular (I can't speak to SUSE) maintains its consistency by having a fairly small repo set. |
| 05:33 | | * Vornicus hasn't met an operating system that isn't made of spiders on some level. |
| 05:33 | < Namegduf> | I'm unfond of Red Hat, but have never used it much. |
| 05:33 | < Vornicus> | However a surprising number of linux distributions start out made of spiders, and have no kittens involved whatsoever. |
| 05:33 | <@McMartin> | I had to compile quite a bit by hand on Fedora but I never had the slightest difficulty having my own stuff coexist with the autoinstalled stuff. |
| 05:33 | <@ToxicFrog> | McMartin: SUSE likewise, but they have an officially sanctioned deployment infrastructure for community repos and even one-off package builds |
| 05:33 | <@McMartin> | yum is hobbled there by being rpm-based~ |
| 05:33 | < Namegduf> | And yeah, I'll readily agree; Debian Stable is for Stable stuff. With a few biting things, and I think to a degree different versions of Python in particular were/are one. |
| 05:34 | < Namegduf> | I know yum is a horrible RAM hog, which makes me go "ew". |
| 05:34 | <@ToxicFrog> | So enabling the equivalent of fedora-extras, or even installing obscure-package-that-one-other-person-ever-wanted-to-build, is pretty much painless. |
| 05:34 | < Namegduf> | That's about all I know. |
| 05:34 | <@McMartin> | TF: fedora-extras is actually gone now; a bunch of the third-party guys merged. |
| 05:34 | <@McMartin> | Once I get my new machine built I'm going to try to convert Zinglon to Fedora 12. |
| 05:34 | < Namegduf> | I find Debian Testing perfectly awesome for the desktop, but I'll admit it's not so awesome there. |
| 05:34 | <@ToxicFrog> | yum is slow, RAM-hungry, and has locking issues, but I find the way the output and commands are structured are more intuitive to me than apt-get is. |
| 05:35 | < Namegduf> | That said, Ubuntu is way too, er, idiot-proof-aiming |
| 05:35 | < Namegduf> | I don't use apt-get at all now, really. |
| 05:35 | < Namegduf> | I always use aptitude. |
| 05:35 | < Vornicus> | Even Windows has kittens in it. |
| 05:35 | < Namegduf> | They're dead kittens, though. |
| 05:35 | <@ToxicFrog> | Windows printing API <3 |
| 05:35 | < Namegduf> | Windows * API is ow |
| 05:35 | < Namegduf> | THat said, I'll believe it could beat printing on *nix. |
| 05:35 | <@McMartin> | I'm a fan of Win32 named pipes. |
| 05:35 | <@ToxicFrog> | The printing API is exactly like drawing to the screen. |
| 05:35 | < Namegduf> | Which seems complicated. |
| 05:36 | <@ToxicFrog> | Except it goes to a printer rather than the framebuffer. |
| 05:36 | < Namegduf> | Yeah, and printing to the screen is Painful. |
| 05:36 | <@McMartin> | TF: Yeah. Apple's is similar. |
| 05:36 | | * Namegduf had some experience there |
| 05:36 | < Vornicus> | Nah, Windows printing is "Okay, well... I know how to draw it on the screen. Can you, like... yeah. thanks" |
| 05:36 | <@ToxicFrog> | In *nix, the printing API, last I checked, was "have your program generate postscript" |
| 05:36 | <@ToxicFrog> | (Although, granted, this was a few years ago) |
| 05:36 | < Vornicus> | Which isn't all that bad a printing API all told, TF. |
| 05:36 | <@McMartin> | Modern OS X, you do the Windows Printing thing, basically, and then the OS turns it into PostScript. |
| 05:36 | < Namegduf> | I wouldn't be surprised if Windows is easier. |
| 05:36 | <@ToxicFrog> | McMartin: my main objection to win32 named pipes is that they aren't named pipes, they're named sockets |
| 05:37 | <@ToxicFrog> | Vornicus: unless your program is already rendering things to the screen. |
| 05:37 | < Namegduf> | Windows's inability to select() on anything but sockets is hilarious |
| 05:37 | < Namegduf> | And by hilarious, I mean crippling |
| 05:37 | <@ToxicFrog> | In which case now you need one view that does X11 calls (or GTK or what have you), and one view that emits postscript |
| 05:37 | < Namegduf> | I had a Python program fail to port to Windows for that reason alone. |
| 05:37 | <@McMartin> | Windows is not POSIX, and pretending it is will lead only to suffering |
| 05:37 | < Vornicus> | Well, yes. |
| 05:37 | < Namegduf> | Yeah. |
| 05:37 | | * Vornicus has never needed to select on anything but sockets. |
| 05:37 | <@McMartin> | (Windows is, in fact, VMS~) |
| 05:38 | <@ToxicFrog> | This also makes it easy to write programs that generate print output that looks totally different than what it looks like on screen! |
| 05:38 | | * Namegduf needed to select on a pipe to a subprocess, AND sockets |
| 05:38 | < Namegduf> | (Was a stupid relay-data-between-the-two scenario) |
| 05:38 | <@ToxicFrog> | UNIX domain sockets! |
| 05:38 | <@ToxicFrog> | oh wait! |
| 05:38 | < Namegduf> | (No idea how you'd sanely do it on Windows, with how select() sucks) |
| 05:39 | <@ToxicFrog> | You don't have those either~ |
| 05:39 | | * Namegduf generally doesn't care to try to support Windows nowadays. |
| 05:39 | <@ToxicFrog> | (you do have named pipesockets, but I'm not sure if they're selectable) |
| 05:39 | < Namegduf> | I'm all for portable code, but Windows needs more than portable, compliant code. |
| 05:39 | <@McMartin> | One thing windows got right but gets no credit for getting right is library APIs that are both wide and byte characters. |
| 05:39 | < Namegduf> | Not a little more, a lot more. |
| 05:39 | < Namegduf> | It's like writing for IE6 |
| 05:40 | <@ToxicFrog> | Windows is remarkably successful at making you choose between "compatible with windows", "compatible with everything but windows", and "takes twice as long to develop" |
| 05:40 | <@McMartin> | I only mention this because wxWidgets failed at it so hilariously hard that I hesitate to even call it a library. |
| 05:40 | < Namegduf> | McMartin: Unless wide means UTF_8, I consider it kinda pointless. |
| 05:40 | <@ToxicFrog> | I tend to solve this by using a HLL with a cross-platform interpreter and libraries, so someone else gets to figure out how to implement ftw() on windows. |
| 05:40 | <@McMartin> | TF: My experience with tri-platform development says... yeah. |
| 05:40 | <@McMartin> | That this is only true for system software. |
| 05:41 | <@McMartin> | And if you want decent behavior, three times as long, at least, because OS X's non-POSIX facilities tend to be enough better that you Ought To Use Them. |
| 05:41 | <@McMartin> | On the other hand, there's one way to do it right on Win32 and OS X and it generally works, while seriously portable POSIX work is a thicket of autoconf edge cases. |
| 05:41 | < Namegduf> | Your fault for using autoshit |
| 05:42 | <@McMartin> | Which is why people tend go "fuck it", use what their machine has, and find that people can't run on Stable. |
| 05:42 | < Namegduf> | I'm not at all sure a worst-designed build system could have been implemented. |
| 05:42 | <@McMartin> | I use makedepend and a handful of templates. |
| 05:42 | < Namegduf> | Also, I've never seen any major dev project totally ignore "which versions of software are available, which should we support" |
| 05:42 | < Namegduf> | And I follow goddamn IRC development |
| 05:42 | <@ToxicFrog> | Hmm. Good: LuaPilot is properly generating stack traces. |
| 05:43 | <@ToxicFrog> | Bad: it shouldn't be crashing~ |
| 05:43 | < Namegduf> | I would honestly say anyone who just develops for their own system is responsible for their software breaking with older versions, not the older versions at fault for existing. |
| 05:43 | < Namegduf> | And the older versions are a lot more predominant. |
| 05:44 | <@McMartin> | There are limits. |
| 05:44 | < Namegduf> | It's kinda like writing for CSS3 then whining that your website doesn't look as good in Firefox as it does in Chrome |
| 05:44 | < Namegduf> | And blaming everyone who uses Firefox. |
| 05:44 | <@McMartin> | UQM doesn't seriously intend to support MacOS 9, even though we have open tickets about why you can't compile for it. |
| 05:44 | < Namegduf> | There *are* limits, but Debian stable stays within them. |
| 05:44 | | * Vornicus still can't figure out the spec for CSS3 or XHTML. |
| 05:44 | <@McMartin> | When I was first developing Blorple, Stable was not within those limits. |
| 05:45 | <@McMartin> | IIRC, I needed to take an image and have it scale to the size of its containing widget. |
| 05:45 | < Namegduf> | If you require 2.5, the software's at fault. |
| 05:45 | | * Namegduf notes his university machines, running Red Hat, are running 2.4 |
| 05:45 | < Namegduf> | *university's |
| 05:45 | < Namegduf> | Well, Red Hat Linux and Solaris, respectively. |
| 05:45 | <@McMartin> | For clarity, RH != Fedora, and hasn't been for quite some time. |
| 05:46 | < Namegduf> | Right. My point is, if you were designing for 2.5 back then, you were basing your stuff on the latest and greatest and had little grounds to expect it to work. |
| 05:46 | < Namegduf> | 2.4 is still in use *now*. |
| 05:46 | < Namegduf> | And then was quite a while ago. |
| 05:46 | <@McMartin> | No, I was designing for Gtk+ 2.1 when Gtk+ was at 2.8 |
| 05:46 | <@McMartin> | And Stable couldn't run it. |
| 05:46 | <@ToxicFrog> | Namegduf: yes, by legacy software that doesn't work with 2.5 or 2.6 and which no-one has bothered updating. |
| 05:47 | < Namegduf> | ToxicFrog: As I said, my university uses it. |
| 05:47 | <@McMartin> | (In the end, the whole damned thing was recoded in Swing, targeting Java 1.5 after 1.6 had been out for some time.) |
| 05:47 | < Vornicus> | Um. |
| 05:47 | < Vornicus> | SOme places still run py152 |
| 05:47 | <@McMartin> | That's a major version change. |
| 05:47 | <@McMartin> | I have no objections to py1 and p2 being separate packages. |
| 05:47 | < Vornicus> | True this. |
| 05:47 | <@McMartin> | Or py3 |
| 05:48 | < Namegduf> | Python 2.5 also doesn't offer that many shiny new features. |
| 05:48 | < Namegduf> | Designing for it is Laziness. |
| 05:48 | <@McMartin> | Given that that py1 and py2 are practically entirely separate languages... |
| 05:48 | < Namegduf> | Really, a complete lack of care to support older systems. |
| 05:48 | <@McMartin> | Namegduf: One of the three cardinal virtues, sir. |
| 05:48 | <@McMartin> | In fact, my last python program requires 2.6 unless you're on a Mac |
| 05:48 | <@McMartin> | Because I'll be god-damned if I'm going to write a complete Plist parsing library instead of a single line of code. |
| 05:48 | < Namegduf> | Cool, but you've no right to complain if it doesn't work for other people usefully. |
| 05:49 | <@McMartin> | The script exists solely to be a frontend to a library added in 2.6. |
| 05:49 | < Namegduf> | Right, sounds sensible, then. |
| 05:49 | <@McMartin> | So yes, I do have a right to complain, viz, what the fuck did they expect. |
| 05:50 | <@McMartin> | (My main OSS project is an odd duck as SDL's developmental relationship with us is rocky and long-term and incestuous) |
| 05:50 | < Namegduf> | My point is that people who use new APIs the month they come out are the ones to yell at for not supporting stable/slowly updating systems, not the failing of stable/slowly updating systems to work with it. |
| 05:50 | <@McMartin> | There's a lot of "we work with any version of SDL except these three which aren't contiguous" |
| 05:50 | <@McMartin> | Yeah, and we have a lot of "OH HEY YOU HAVE A BUG" to which we reply "that bug is in upstream code that was fixed and published last year, get lost" |
| 05:51 | <@McMartin> | I'm still not entirely sure what exactly it is UQM does that makes SDL hit weird corner cases, but boy howdy does it. |
| 05:51 | < Namegduf> | I'm still not sure what business anyone has installing GTK+ on Debian Stable, though, haha. |
| 05:51 | | * Namegduf would not run a desktop off it. |
| 05:52 | <@McMartin> | Enough people did that I got bug reports~ |
| 05:52 | <@McMartin> | (And then you have Apple, which adds core functionality that should never have been missing extremely late in the game.) |
| 05:52 | <@McMartin> | (It wasn't until 10.4 that you could programmatically set duplex mode) |
| 05:52 | < Namegduf> | Cases of that are reasonable. |
| 05:53 | < Namegduf> | There's "avoiding using tiny convieniences and breaking things for the sake of it" |
| 05:53 | < Namegduf> | And there's "using major new things which should never have been missing" |
| 05:53 | < Namegduf> | I'd still think hard before dropping previous version support, though. |
| 05:53 | <@McMartin> | I suspect the real error in Stable was putting 2.0 in, since it was .0 |
| 05:54 | <@McMartin> | The first version of Gtk+ that wasn't completely inadaquate for everything was 2.4 or so |
| 05:54 | < Namegduf> | That's possible. |
| 05:54 | | * Namegduf has no idea how long ago this must have been |
| 05:54 | < Namegduf> | I *guess* it might have been Etch, if GTK+ moved up 12 versions in 18 months. |
| 05:54 | < Namegduf> | Otherwise, it'd need to be the one before, I think. |
| 05:54 | <@McMartin> | ... what was th eone before? |
| 05:55 | <@McMartin> | IIRC it was Etch that he wanted to not upgrade *to*. |
| 05:55 | < Namegduf> | I don't remember, too far back. |
| 05:55 | <@McMartin> | Etch had *just* come out. |
| 05:55 | < Namegduf> | Ah. |
| 05:55 | < Namegduf> | Silly person. |
| 05:55 | < Namegduf> | In that case, you're perfectly fine not to support Oldstable |
| 05:55 | < Namegduf> | Which is, as the name implies, Old |
| 05:55 | <@McMartin> | As it turned out, Gtk was the wrong solution anyway. |
| 05:56 | | * McMartin eventually recoded it in Swing, and sad to say, this was far easier than PyGtk. |
| 05:56 | < Namegduf> | I'll need to use PyGtk sometime. |
| 05:56 | <@McMartin> | Glade exists now. |
| 05:56 | < Namegduf> | Right now, I need to work out, given a binary, how to emulate an RNG |
| 05:56 | < Namegduf> | No source. |
| 05:56 | < Namegduf> | I'm debating disassembling it, but that might be cheating. |
| 05:57 | <@McMartin> | "cheating"? |
| 05:57 | < Namegduf> | University thing. |
| 05:57 | < Namegduf> | It's deterministic or has a fixed seed. |
| 05:57 | <@McMartin> | Oh |
| 05:57 | < Vornicus> | If you can grab a large run from the RNG, you should be able to come up with a command to battle it. |
| 05:58 | < Vornicus> | Indeed - for lcg generators you can do it, usually, given three values. |
| 05:58 | < Vornicus> | mh, four. |
| 05:59 | | * McMartin consults |
| 05:59 | <@McMartin> | Aha |
| 05:59 | <@McMartin> | This was in fact just *before* Etch became Stable. |
| 05:59 | < Namegduf> | Ah. |
| 05:59 | < Namegduf> | Stable at its oldest. |
| 05:59 | <@McMartin> | Which was probably also why it was so ludicrously out of date. |
| 06:00 | <@McMartin> | (2 years, 6 months ago. Thank you, SF SVN Browse.) |
| 06:03 | <@McMartin> | (Though that was during initial import.) |
| 06:05 | <@McMartin> | Speaking of which, there were some semi-core things added in Java 6 I didn't use in that program later on. |
| 06:05 | <@McMartin> | It's probably been out long enough that I can demand Java 6 now. |
| 06:05 | <@McMartin> | (in particular, proper sort-by-columns behavior) |
| 06:06 | | Derakon is now known as Derakon[AFK] |
| 06:38 | | AnnoDomini [farkoff@Nightstar-189b1477.adsl.tpnet.pl] has joined #code |
| 07:10 | | ErikMesoy|sleep is now known as ErikMesoy |
| 07:27 | | Reivles [reaverta@ServerAdministrator.Nightstar.Net] has quit [Ping timeout: 121 seconds] |
| 07:27 | | Reivles [reaverta@ServerAdministrator.Nightstar.Net] has joined #code |
| 07:59 | | Rhamphoryncus [rhamph@Nightstar-a62bd960.abhsia.telus.net] has quit [Client exited] |
| 08:39 | | Vornicus is now known as Vornicus-Latens |
| 08:47 | | Tarinaky [Tarinaky@Nightstar-2ee820d2.adsl.virginmedia.net] has quit [Connection closed] |
| 09:00 | | MyCatVerbs [mycatverbs@Nightstar-f43ca811.blueyonder.co.uk] has quit [Connection closed] |
| 10:42 | | MyCatVerbs [mycatverbs@Nightstar-492e7e67.co.uk] has joined #code |
| 11:11 | | AnnoDomini [farkoff@Nightstar-189b1477.adsl.tpnet.pl] has quit [Ping timeout: 121 seconds] |
| 11:14 | | AnnoDomini [farkoff@Nightstar-a682c050.adsl.tpnet.pl] has joined #code |
| 11:17 | | You're now known as TheWatcher |
| 12:43 | | Attilla [The.Attilla@FBC920.A5C359.15BD32.425F3F] has joined #code |
| 13:33 | | gnolam [lenin@Nightstar-38637aa0.priv.bahnhof.se] has joined #code |
| 15:06 | | Syloqs_AFH [Syloq@NetworkAdministrator.Nightstar.Net] has joined #code |
| 15:07 | | Syloqs_AFH is now known as Syloqs-AFH |
| 16:00 | | Rhamphoryncus [rhamph@Nightstar-a62bd960.abhsia.telus.net] has joined #code |
| 17:36 | | Attilla [The.Attilla@FBC920.A5C359.15BD32.425F3F] has quit [Ping timeout: 121 seconds] |
| 17:45 | | * gnolam rides the refactor tractor. |
| 17:56 | < gnolam> | o/` Wer is überall der Erste? |
| 17:56 | < gnolam> | Das ist Fritz, der Refaktorist! o/` |
| 17:59 | | Attilla [The.Attilla@FBC920.92E3D3.6059E7.06CE27] has joined #code |
| 18:09 | | You're now known as TheWatcher[afk] |
| 18:40 | | Tarinaky [Tarinaky@Nightstar-a54b1e17.adsl.virginmedia.net] has joined #code |
| 18:44 | < SmithKurosaki> | Really now gnolam? |
| 19:24 | | crem [moo@Nightstar-8ca3eea7.adsl.mgts.by] has quit [Client closed the connection] |
| 19:31 | | crem [moo@Nightstar-8ca3eea7.adsl.mgts.by] has joined #code |
| 19:52 | | You're now known as TheWatcher |
| 20:26 | | Syloqs-AFH [Syloq@NetworkAdministrator.Nightstar.Net] has quit [Ping timeout: 121 seconds] |
| 20:34 | < gnolam> | Indeed. |
| 20:46 | < SmithKurosaki> | AhiShocking |
| 21:40 | | Syloqs_AFH [Syloq@NetworkAdministrator.Nightstar.Net] has joined #code |
| 21:41 | | Syloqs_AFH is now known as Syloqs-AFH |
| 21:43 | | Derakon[work] [Derakon@Nightstar-d44d635e.ucsf.edu] has joined #code |
| 21:43 | | * Derakon[work] eyes these 51 Python files, starts thinking about how they should be properly organized. |
| 21:45 | <@McMartin> | 1 bug; 6 hours of work; 2-line fix. |
| 21:48 | < Vornicus-Latens> | All too familiar. |
| 21:48 | < Derakon[work]> | I've been there. |
| 21:49 | < Derakon[work]> | Hell, I wouldn't be surprised if this crash I'm trying to figure out has a short fix, and I've been working on it off and on for weeks now. |
| 21:56 | | * Derakon[work] creates the directories "dialogs", "experiment", and "ui", is left with 29 files to classify. |
| 22:33 | < AnnoDomini> | Don't forget to make the 'PETE' catalogue. |
| 22:34 | | * Derakon[work] facepalms. |
| 22:34 | <@McMartin> | PETE? |
| 22:34 | < Derakon[work]> | The original screenshot of the UI has a button in red labeled "Pete". |
| 22:34 | < Derakon[work]> | This is the switch-user button, which for whatever reason used to use as its label the current username. |
| 22:36 | <@TheWatcher> | The terrifying thing is that I can see the logic behind that |
| 22:37 | < Derakon[work]> | On the one hand, I admit that the new UI doesn't tell you who you're logged in as. |
| 22:37 | < Derakon[work]> | On the other hand, this is not generally a problem. |
| 22:37 | <@TheWatcher> | On the gripping hand, shouldn't that be up to windows anyway? Or do you do user wrangling inside the app? |
| 22:38 | < Derakon[work]> | This is used primarily to automatically determine where experiment data should be saved. |
| 22:38 | <@TheWatcher> | Ah |
| 22:38 | < Derakon[work]> | It's not an XP login. |
| 22:38 | <@TheWatcher> | Righto |
| 22:38 | < Derakon[work]> | In fact, the userlist used to be hardcoded; these days it's generated based on the directories in the A_OMX_DATA directory on the data drive. |
| 22:41 | | * gnolam arrrrghs. |
| 22:42 | | * gnolam facepalms, kicks self. |
| 22:42 | | AkhlD [akhld@6CF571.5B5DAE.62FF1A.119D89] has joined #code |
| 22:43 | | AkhlD [akhld@6CF571.5B5DAE.62FF1A.119D89] has left #code ["Leaving"] |
| 22:43 | <@TheWatcher> | gnolam: What's up? |
| 22:46 | < gnolam> | I've been trying to refactor this class for a couple of hours now. For some reason, removing a now obsolete function call led to incorrect results. |
| 22:47 | < gnolam> | So I've spent the best part of an hour trying to figure out just how that could be - because on the face of it, what was happening was /impossible/. |
| 22:47 | < gnolam> | Somehow, a const function was modifying state. |
| 22:48 | <@TheWatcher> | ... |
| 22:49 | | Derakon[work] [Derakon@Nightstar-d44d635e.ucsf.edu] has quit [[NS] Quit: Leaving] |
| 22:51 | <@TheWatcher> | So what was really happening? |
| 22:52 | < gnolam> | It turned out to be a bloody name clash. :P |
| 22:52 | <@TheWatcher> | ...ouch |
| 22:52 | <@TheWatcher> | >.< |
| 22:53 | < gnolam> | Turns out the function wasn't completely obsolete just yet. The result of its calculation /was/ in fact used a bit further down. |
| 22:54 | < gnolam> | And the class had a member variable whose existence I had forgotten completely... with the same name as one of the "temporary" variables used to store the removed function's calculation. |
| 22:56 | < gnolam> | So instead of getting an undeclared identifier error when the function and variables were removed, which would have alerted me to the need for some more changes, it happily continued with the calculations, with one term having an off-the-walls wrong value. |
| 22:56 | < gnolam> | So... so... stupid. |
| 22:57 | <@TheWatcher> | Easy done though :/ |
| --- Log closed Sat Oct 31 00:00:19 2009 |