| --- Log opened Mon Apr 28 00:00:14 2008 |
| 00:23 | <@Vornicus> | arg what the crap |
| 00:23 | | * Vornicus shoots git. |
| 00:23 | <@Vornicus> | Whyyyyyy don't you tell me the requirements I fail /before/ spending an age building. |
| 00:24 | <+Moltare> | all programming languages are gits :/ |
| 00:36 | | You're now known as TheWatcher[T-2] |
| 00:38 | | gnolam [lenin@Nightstar-10613.8.5.253.static.se.wasadata.net] has quit [Quit: Z?] |
| 00:39 | | You're now known as TheWatcher[zZzZ] |
| 00:50 | | * ToxicFrog blinks at Vorn |
| 00:51 | <@Vornicus> | it took ten minutes and then complained about not being able to find some random perl header. |
| 00:51 | | Vornicus is now known as Finerty |
| 01:27 | <@McMartin> | Vorn: Are you building git or using it for something else? |
| 01:27 | <@Finerty> | Building. |
| 02:07 | | Shoukanjuu [~Shoukanju@Nightstar-18016.dhcp.embarqhsd.net] has quit [Quit: You're gonna carry that weight.] |
| 02:07 | | Shoukanjuu [~Shoukanju@Nightstar-18016.dhcp.embarqhsd.net] has joined #code |
| 02:27 | | Moltare is now known as Molgorn |
| 02:43 | | Reiv [~reaverta@Admin.Nightstar.Net] has joined #Code |
| 02:43 | | RBot [~Reiver@Nightstar-22337.xdsl.xnet.co.nz] has joined #Code |
| 02:44 | | DiceBot [~Reiver@Nightstar-11590.xdsl.xnet.co.nz] has quit [Ping Timeout] |
| 02:44 | | RBot is now known as DiceBot |
| 02:45 | | Reiver [~reaverta@Admin.Nightstar.Net] has quit [Ping Timeout] |
| 02:45 | | Reiv is now known as Reiver |
| 04:15 | | Chalcedon [~Chalcy@Nightstar-488.ue.woosh.co.nz] has joined #code |
| 04:15 | | mode/#code [+o Chalcedon] by ChanServ |
| 04:32 | | JeffL [~JPL@Nightstar-12594.dsl.sndg02.pacbell.net] has quit [Ping Timeout] |
| 04:55 | | Finerty is now known as Vornicus |
| 06:34 | | AnnoDomini [AnnoDomini@Nightstar-29716.neoplus.adsl.tpnet.pl] has joined #Code |
| 06:34 | | mode/#code [+o AnnoDomini] by ChanServ |
| 07:42 | | Vornicus is now known as Vornicus-Latens |
| 08:43 | | You're now known as TheWatcher |
| 09:18 | | Thaqui [~Thaqui@Nightstar-123.jetstream.xtra.co.nz] has joined #code |
| 09:18 | | mode/#code [+o Thaqui] by ChanServ |
| 10:01 | | Chalcedon [~Chalcy@Nightstar-488.ue.woosh.co.nz] has quit [Quit: Leaving] |
| 10:10 | | AnnoDomini [AnnoDomini@Nightstar-29716.neoplus.adsl.tpnet.pl] has quit [Ping Timeout] |
| 10:17 | | AnnoDomini [AnnoDomini@Nightstar-29227.neoplus.adsl.tpnet.pl] has joined #Code |
| 10:17 | | mode/#code [+o AnnoDomini] by ChanServ |
| 12:03 | | Thaqui [~Thaqui@Nightstar-123.jetstream.xtra.co.nz] has left #code [Leaving] |
| 12:06 | | gnolam [lenin@Nightstar-10613.8.5.253.static.se.wasadata.net] has joined #Code |
| 12:06 | | mode/#code [+o gnolam] by ChanServ |
| 12:07 | | C_tiger [~c_wyz@Nightstar-18112.nycmny.east.verizon.net] has quit [Ping Timeout] |
| 13:07 | | Brother_Willibald [lenin@Nightstar-10613.8.5.253.static.se.wasadata.net] has joined #Code |
| 13:07 | | gnolam [lenin@Nightstar-10613.8.5.253.static.se.wasadata.net] has quit [Ping Timeout] |
| 13:18 | <@AnnoDomini> | This is quite odd. I'm trying to do the LBM drawing thing, as McMartin suggested, and I'm getting weird things. If I want the bloody thing to match 320x200 resolution, so that where I point and where the dot is put are the same, dots appear only if the mouse movement is fast enough. |
| 13:24 | | Brother_Willibald is now known as gnolam |
| 13:31 | <@AnnoDomini> | I don't know what the hell is happening. When I use the incorrect value for X, being twice what is needed, it works after a fashion, but the drawing is smooth enough. If I put "SAR CX,1" to divide it in half, it fails to work mostly at all, giving results only when mouse movement is high. |
| 13:42 | <@AnnoDomini> | http://pastie.caboo.se/188014 <- Could anyone with TASM and DosBox tell me if this works as crappily for them as for me? |
| 13:47 | <@AnnoDomini> | It similarly refuses to work as I want it to if I set the resolution to 640x200. |
| 13:55 | <@AnnoDomini> | Could it be that the cursor somehow deletes what is underneath it? |
| 13:57 | <@AnnoDomini> | Hrmph. It is! |
| 14:15 | <@AnnoDomini> | Haha. Success. |
| 15:14 | <@ToxicFrog> | Related thought: could it be relative mouse movement? |
| 15:21 | <@AnnoDomini> | What? |
| 15:24 | <@AnnoDomini> | The coordinates I get with INT 33,3 are related to point 0,0 in the upper left corner. But they're always of the ranges 0..639 and 0..199, regardless of video mode. |
| 15:39 | <@McMartin> | That sounds like a hack for textmode mice, honestly |
| 15:40 | <@McMartin> | 640x200 is the highest res VGA supported. |
| 15:40 | <@McMartin> | So, one mouse driver, all modes. |
| 15:41 | <@McMartin> | It's also possible that these drivers weren't intended to be polled. |
| 15:41 | <@McMartin> | That you are instead suppose to be reacting to interrupts if you want to do drag |
| 15:41 | <@McMartin> | The easiest thing now would be to place a stamp at wherever the mouse is clicked. |
| 15:41 | <@McMartin> | I suggest finding an image of a bullet hole in glass~ |
| 15:41 | <@McMartin> | BLAM BLAM BLAM |
| 15:45 | <@ToxicFrog> | I thought 640x480x4 was the highest VGA mode. |
| 15:46 | <@McMartin> | Not standard until after Windows 95. |
| 15:46 | <@McMartin> | 640x480 was firmly in the "SVGA" bin in the early 90s, at least. |
| 15:47 | <@ToxicFrog> | Aah. |
| 15:47 | <@McMartin> | Though the only cases I can think of of anyone using it are all 640x480x16. |
| 15:49 | <@ToxicFrog> | 16 bits, or 16 colors? |
| 15:49 | <@ToxicFrog> | I'm using bits. |
| 15:52 | <@McMartin> | Oh. |
| 15:52 | <@McMartin> | Yeah, colors. |
| 15:52 | <@McMartin> | Because the MCGA-compatible modes were always 320x200x256 |
| 15:53 | <@ToxicFrog> | Aah. |
| 15:53 | | * ToxicFrog generally uses (and sees) pixels x pixels x bits @ Hz |
| 15:55 | <@McMartin> | Yeah. I tend to treat pre-Win32 machines as being in the C64 era~ |
| 17:17 | <@jerith> | So, quick HTML question. |
| 17:17 | <@jerith> | Is it legal and advisable to have <a name="foo" href="#bar">something</a>? |
| 17:18 | <@jerith> | Basically, I want to add a footnote to my LJ entry. |
| 17:18 | <@ToxicFrog> | As far as I'm aware, yes, but test it and find out. |
| 17:18 | <@ToxicFrog> | The idea being that the footnote itself will link to #foo? |
| 17:19 | <@McMartin> | IFAIK this is How I'ts Done, but yeah try it out. |
| 17:20 | <@jerith> | The alternative is <a name="foo" /><a href="#bar">something</a> |
| 17:20 | <@jerith> | I want the footnote and the number to link back to each other. |
| 17:20 | <@jerith> | HTML has a <super>foo</super> tag, right? |
| 17:21 | <@jerith> | Ah, <sup>. |
| 17:22 | <@AnnoDomini> | <sup>, <dawg>? |
| 17:22 | | * jerith sends AnnoDomini to the Bad Place for that. |
| 17:43 | <@AnnoDomini> | Forever? |
| --- Log closed Mon Apr 28 17:43:35 2008 |
| --- Log opened Mon Apr 28 17:43:40 2008 |
| 17:43 | | TheWatcher [~chris@Nightstar-29731.dsl.in-addr.zen.co.uk] has joined #code |
| 17:43 | | Irssi: #code: Total of 20 nicks [13 ops, 0 halfops, 1 voices, 6 normal] |
| 17:43 | | mode/#code [+o TheWatcher] by ChanServ |
| 17:44 | | Irssi: Join to #code was synced in 52 secs |
| 18:23 | | AnnoDomini is now known as Steven |
| 18:23 | | Steven is now known as AbuDhabi |
| 18:23 | | AbuDhabi is now known as Steve |
| 18:47 | | C_tiger [~c_wyz@96.232.28.ns-12390] has joined #code |
| 18:47 | | mode/#code [+o C_tiger] by ChanServ |
| 20:18 | | * McMartin fires up Fedora 9 Preview Release. |
| 20:35 | <@jerith> | Is the Star Trek Movie rule still in effect with Fedora? |
| 20:35 | <@jerith> | Or was that only when they still had "core"? |
| 20:36 | <@McMartin> | FC5 was good. |
| 20:37 | <@McMartin> | The last utterly terrible Fedora was FC4. |
| 20:37 | <@McMartin> | But pre-5, only 3 was usable, so. |
| 20:37 | <@McMartin> | So I guess I'm wondering which star trek rule you mean~ |
| 20:37 | <@McMartin> | One can't deduce much from a liveCD on an ancient laptop, though. |
| 20:38 | <@McMartin> | It looks like it more or less works, but I didn't get the KDE one. |
| 20:38 | <@McMartin> | And since F9 is part of the KDE4 rollout, I imagine that will be Significant. |
| 20:38 | <@jerith> | Oh, 'twas an inverse Star Trek Movie rule. |
| 20:38 | <@McMartin> | And yeah, that rule is gone now. |
| 20:39 | <@McMartin> | I've used every Fedora release but 2 and 4, and all the ones from 5+ are actually recommendable. |
| 20:39 | <@McMartin> | 3 was usable but not recommendable. |
| 20:39 | <@jerith> | Ah. |
| 20:39 | <@McMartin> | 7 was when yum stopped sucking. |
| 20:39 | <@jerith> | I'm on RHEL these days. |
| 20:39 | <@jerith> | yum stopped sucking? |
| 20:40 | <@jerith> | I didn't realise this was possible. |
| 20:40 | <@McMartin> | It both got better about keeping its cache coherent, and it got much faster about sorting out dependencies. |
| 20:40 | <@jerith> | Does it still require you to be online to search the package list? |
| 20:40 | <@McMartin> | Attempting to simultaneously install x86_64 and i386 versions of apps would occasionally confuse it. |
| 20:40 | <@jerith> | Does it still refuse to let you search said list while another instance is running? |
| 20:41 | <@McMartin> | I haven't tested either of those things. |
| 20:41 | <@jerith> | Does it still suck at downgrading packages? |
| 20:41 | <@McMartin> | Indeed, the first would be difficult to test without etherial. |
| 20:41 | <@McMartin> | So, my requirements are "does it correctly resolve dependencies", "does it not pull in 500MB of unrelated bullshit", and "does it complete in minutes". |
| 20:42 | <@McMartin> | Of yum, apt, and emerge, yum is the oonly one that's managed all three for me. |
| 20:42 | <@jerith> | The first is because the one I'm using *requires* one to update the repo metadata on startup. |
| 20:42 | <@McMartin> | That is news to me. |
| 20:42 | <@jerith> | You can force -C to work offline, but then some other things break. |
| 20:42 | <@McMartin> | I suspect that if one has not cleared out the metadata with yum clean all that it doesn't touch the network. |
| 20:42 | <@McMartin> | I haven't tried simultaneous searches. |
| 20:43 | <@jerith> | No, yum lacks a separate "update metadata" step. Thus, it updates metadata from the network whenever it thinks it needs to. |
| 20:44 | <@McMartin> | And if it fails to do so, it then refuses to run? |
| 20:44 | <@McMartin> | Because I haven't run into that since FC1. |
| 20:44 | <@McMartin> | My usual problem is it taking 40 minutes to parse the package list. |
| 20:44 | <@McMartin> | And that stopped in 7. |
| 20:45 | <@jerith> | I've stopped running it without a net connection. |
| 20:45 | <@jerith> | There is also no way to tell it to download packages, but not update repo metadata. |
| 20:45 | <@McMartin> | How is this a feature? |
| 20:45 | | C_tiger [~c_wyz@96.232.28.ns-12390] has quit [Ping Timeout] |
| 20:45 | <@jerith> | Which is useful if you know there's a bunch of stuff that wants updating but don't want to pull it all in now. |
| 20:46 | <@McMartin> | Er, wouldn't you want the reverse for that? |
| 20:46 | <@jerith> | McMartin: It's not, unless bandwidth is a scarce and expensive commodity. |
| 20:46 | <@jerith> | If I haven |
| 20:46 | <@McMartin> | If the problem is "I can't look up the latest version of *one* package without doing so for *all* of them", how does that differ from anyone else? |
| 20:47 | <@jerith> | If I haven't updated in a month or so, but need to install a random Python module, I don't want it to suck down new repo metadata and then try to update the rest of my system. |
| 20:47 | <@McMartin> | Um |
| 20:47 | <@McMartin> | "yum install" does not force "yum upgrade" and TMK never has. |
| 20:47 | <@jerith> | Correct. |
| 20:47 | <@McMartin> | So it doesn't do what you're complaining about? |
| 20:48 | <@jerith> | But if the new version of whatever it is I want to install has dependencies that are new versions... |
| 20:48 | <@jerith> | I have been bitten by this in the past. |
| 20:48 | <@McMartin> | ... then you can't install it anyway. |
| 20:48 | <@McMartin> | gentoo would cheerfully attempt to compile stuff that won't work, and wreck your entire system. |
| 20:48 | <@jerith> | I could install the old version. |
| 20:49 | <@McMartin> | OK, so I actually consider this a feature. |
| 20:49 | <@McMartin> | I've been bitten enough by APT doing the reverse that I abandoned it. |
| 20:49 | <@jerith> | In yum, installing an old version of something is difficult to the point where I gave up on it and rebuilt repos. |
| 20:49 | <@McMartin> | "Oh, you actually needed to add all these extra numbers to the end to get the actual correct version of it. Otherwise it just installs something heinously obsolete." |
| 20:49 | <@jerith> | Typically, I consider "I want to install foo" and "I want to update my system" to be different things. |
| 20:50 | <@jerith> | I always want to install the newest version in my local cache of the repo metadata. |
| 20:50 | <@McMartin> | I'm still trying to figure out how a python package will force the update of anything. |
| 20:50 | <@jerith> | That was a random example. |
| 20:50 | <@jerith> | MAke it an OpenOffice plugin, then. |
| 20:50 | <@jerith> | The old version requires the OOo I have. |
| 20:51 | <@jerith> | The new version requires the next point release. |
| 20:51 | <@McMartin> | Yeah |
| 20:51 | <@jerith> | I don't care about the new feature that requires the new version. |
| 20:51 | <@McMartin> | To date, nobody does that right. =P |
| 20:51 | <@jerith> | I *do* care that OOo is a massive download I can't afford. |
| 20:51 | <@jerith> | I've never had a problem with it in apt. |
| 20:51 | <@McMartin> | Yes. |
| 20:51 | <@jerith> | Sometimes if it's very old it'll tell me it can't find the .deb. |
| 20:52 | <@McMartin> | Because in APT if you want to upgrade you have to install an entirely new set of packages and automatic upgrades simply don't happen at all. |
| 20:52 | <@McMartin> | And your first attempt at installing it was actually 12 versions too old. |
| 20:52 | <@jerith> | Um. |
| 20:52 | <@jerith> | I do "aptitude update" and "aptitude upgrade" and everything's happy. |
| 20:52 | <@McMartin> | (Still bitter about how hard it was to get a version of GTK+ that wasn't 2.0 when 2.12 was out.) |
| 20:53 | <@jerith> | As long as I do this on a regular basis, my world is happy. |
| 20:53 | <@jerith> | Was that not perhaps a repo issue? |
| 20:53 | <@McMartin> | The only way I could get that consistently was by repeatedly reinstalling successive versions of Ubuntu. |
| 20:53 | <@jerith> | Ah. If you were running an old Ubuntu you were using old repos. |
| 20:54 | <@jerith> | That's like complaining that you can't install KDE4 in FC3. |
| 20:54 | <@McMartin> | My limited experience with Raw Debian made it even worse. |
| 20:54 | | C_tiger [~c_wyz@Nightstar-25834.nycmny.east.verizon.net] has joined #code |
| 20:54 | | mode/#code [+o C_tiger] by ChanServ |
| 20:54 | <@McMartin> | Because if I didn't know precisely the name of the package I wanted to install, I was hosed. |
| 20:54 | <@McMartin> | And while that would normally be OK, when there are 14 million versions of the same package in the repo, it takes a while to find the ones that all play right |
| 20:55 | <@McMartin> | And god help you if you misguessed one of the apps, because now you're going to have multiple *copies* of everything. |
| 20:55 | <@jerith> | Ugh. |
| 20:55 | <@jerith> | Yeah, apt-get sucks somewhat. aptitude is much nicer. |
| 20:56 | <@McMartin> | And the last time I looked at emerge, you could end up *needing* multiple copies side by side but it wouldn't *let you have them* unless you planned it out in advance. |
| 20:56 | <@McMartin> | Which meant you'd get incomprehensible compiler errors an hour after you started your update. |
| 20:56 | <@jerith> | Really? |
| 20:56 | <@McMartin> | USE flags control compilation, but weren't preserved |
| 20:56 | <@jerith> | Last time I checked you could have multiple versions side-by-side if they were configured for different slots. |
| 20:57 | <@McMartin> | Yes. |
| 20:57 | <@McMartin> | But you needed to make slotted versions to begin with. |
| 20:57 | <@jerith> | If they're not, then the package is broken. |
| 20:57 | <@McMartin> | If your system started out unslotted, lolz@u |
| 20:57 | | * McMartin hasn't used gentoo since 2003. |
| 20:57 | <@jerith> | Slots are a function of an ebuild. |
| 20:57 | <@jerith> | apache has 1.3 and 2.x in different slots, for example. |
| 20:57 | <@McMartin> | Yes. When I was first using it, no packages had slots because emerge didn't support it. |
| 20:58 | <@jerith> | Ah. |
| 20:58 | <@McMartin> | I stopped using gentoo because I got sick of everything in the universe breaking whenever there was a major update to something. |
| 20:58 | <@McMartin> | "Oh, I'm sorry, you have all this stuff that needs to be upgraded, but thanks to package blocking, you can't get there from here." |
| 20:58 | <@jerith> | I don't really have a problem with portage, although you can only love it if you really like compiling stuff. |
| 20:58 | <@jerith> | The quality of the repos has always been spotty and got really bad a few years ago. |
| 21:00 | <@jerith> | I see Gentoo as a failed experiment that had a lot of cool stuff happening in it. |
| 21:00 | <@McMartin> | I actually did like Portage as long as I never had to reconfigure anything. |
| 21:00 | <@McMartin> | Once I did, all was lost and I might as well have reinstalled |
| 21:00 | <@McMartin> | Of course, some of this is applications being fundamentally broken. |
| 21:01 | <@McMartin> | wxWidgets, for instance, didn't have slots. |
| 21:01 | <@jerith> | revdep-rebuild was cool, but there was always something that needed hand-fixing. |
| 21:01 | <@McMartin> | But any given application will only compile against it depending on one #define. |
| 21:01 | <@McMartin> | revdep-rebuild should never have been necessary in the first place. |
| 21:01 | <@McMartin> | And it also wasn't documented for years. |
| 21:01 | <@McMartin> | I found out about it by pestering people on IRC. |
| 21:03 | <@jerith> | It's necessary for the case where USE flags or something has changed. It's an expensive operation you don't want to run on every update. |
| 21:03 | <@McMartin> | I found I had to run it any time I upgraded any library. |
| 21:03 | | Chalcedon [~Chalcy@Nightstar-488.ue.woosh.co.nz] has joined #code |
| 21:03 | | mode/#code [+o Chalcedon] by ChanServ |
| 21:03 | <@McMartin> | At minimum, packages in a repo should report their own reverse dependencies, and then have that target what needs to be checked. |
| 21:04 | <@McMartin> | In a binary-based repo, of course, when you update a lib, you also make new binaries for things that depend on it if they break, so someone doing an update never notices that this happened. |
| 21:04 | <@McMartin> | But I generally had to do a revdep-rebuild after every upgrade. |
| 21:04 | <@ToxicFrog> | If by "reverse dependencies" you mean "other packages that depend on this one", this strikes me as almost unworkable, and furthermore unecessary |
| 21:04 | <@ToxicFrog> | Since the package manager can just traverse the list of installed packages and see which ones list the package being updated as dependencies |
| 21:05 | <@McMartin> | TF: Well, I got burned by it constantly in gentoo |
| 21:05 | <@McMartin> | And it actually scanned every ELF-format file to see what it linked and rebuild them if they were suddenly broken |
| 21:05 | <@McMartin> | Which happened Every. Fucking. Time. |
| 21:05 | <@ToxicFrog> | What does tracking reverse dependencies here get you that tracking forward ones doesn't, though |
| 21:06 | <@McMartin> | You could deduce it from forward ones if they weren't a cavern of lies, yes. |
| 21:06 | <@jerith> | portage did the dep traversal. |
| 21:06 | <@McMartin> | Or one could, oh, I don't know, set up one's libraries so that updating them doesn't break apps linked against them |
| 21:07 | <@ToxicFrog> | To me it sounds like the solution here is "fix the forward dependencies, or better yet, cause rebuilding a library to not break stuff that depends on it", not "introduce another set of dependencies, and thus, anothing thing can easily get out of sync with reality and break everything" |
| 21:07 | <@jerith> | McMartin: A lot of the fault there lies with the people who didn't report their deps properly. |
| 21:07 | <@jerith> | And that /could/ have been fixed with a proper sandboxed build/test environment. |
| 21:07 | <@McMartin> | jerith: It actually occurs to me that some of this may be why yum is so inflexible about this. |
| 21:07 | <@jerith> | Which, of course, requires lots of crunch. |
| 21:08 | <@McMartin> | My travails with gentoo definitely make me more willing to accept an inflexible but actually functional package manager. |
| 21:08 | <@jerith> | My travails with Gentoo make me more willing to choose a distro with a quality repo over anything else. |
| 21:10 | <@jerith> | It would actually be quite trivial to write a brute-force repo quality tester, assuming reasonable functional tests for packages. |
| 21:10 | <@McMartin> | Also, I just checked, and yum search and yum update can in fact run concurrently, though if one of them is updating the metadata the other will block until that's done. |
| 21:11 | <@McMartin> | I think RH has some kind of brute-force tester based on their bugzilla |
| 21:11 | <@McMartin> | Which has autofiled bug reports of the form "massbuild failed on package X with new gcc package" and such |
| 21:11 | <@jerith> | For each package, fire up a VM with a minimal system on it and install just that package and its deps. If anything breaks or the tests fail, scream blue murder and mark the package as broken. |
| 21:12 | <@jerith> | McMartin: That must be fairly new, then. I have been trained to wait until yum is finished before I launch a new instance. |
| 21:12 | <@ToxicFrog> | You don't even need a VM, I think, you just need a chroot |
| 21:13 | <@jerith> | Yeah. |
| 21:13 | <@jerith> | Every update to a package queues one of these runs for the new version. |
| 21:14 | <@McMartin> | jerith: As I said initially, my guess is that this was probably from around F7, since that's when I noticed it was running much faster. |
| 21:14 | | * jerith nods. |
| 21:14 | <@jerith> | [jerith@jerith-lap1 ~]$ yum --version |
| 21:14 | <@jerith> | 3.2.8 |
| 21:14 | <@McMartin> | Hrm. That's what mine reports too |
| 21:14 | <@jerith> | I did notice "yum search" results look different now. |
| 21:14 | <@jerith> | This is since I updated to the RHEL5.2 beta. |
| 21:17 | <@jerith> | Ah, that's the other thing I want./ |
| 21:17 | <@jerith> | An easy way to tell which repo you got something out of. |
| 21:19 | <@McMartin> | yum list almost manages this. |
| 21:19 | <@ToxicFrog> | yum list tells you what repos you will get something out of, which is better than apt manages. |
| 21:20 | <@McMartin> | TF: Right, but if it's in core, updates, and installed, it won't tell you if the installed one is from core or updates. |
| 21:20 | <@ToxicFrog> | I don't think the info about what repo you got stuff out of is preserved once it's installed at present, though, so the best it would be able to do (if it did it, which it didn't) would be to say "installed, and you could have gotten it from a or b or c" |
| 21:20 | <@jerith> | yum search tells me "installed" for everything I have installed. |
| 21:20 | <@McMartin> | list, not search. |
| 21:20 | <@McMartin> | And mine lists both "installed" after the repos it was in. |
| 21:21 | <@McMartin> | ... although it doesn't do that on gouda. |
| 21:21 | <@McMartin> | (Which is CentOS, and also only using yum 2.4.3) |
| 21:22 | <@jerith> | Anyways, back to my installing-old-stuff thing. |
| 21:22 | <@ToxicFrog> | "list" and "info" will both say "installed" for anything you have installed, and the repo it would be installed from otherwise. |
| 21:22 | <@ToxicFrog> | "search" doesn't say anything. |
| 21:22 | <@jerith> | It's not useful if bandwidth is effectively infinite and free. |
| 21:22 | <@ToxicFrog> | Isn't it just # yum install package-1.2.3.4? |
| 21:22 | <@jerith> | search used to return the info from enything it matched. |
| 21:23 | <@jerith> | ToxicFrog: When I tried that it yelled at me. |
| 21:23 | <@jerith> | In this case, though, I'm complaining about the equivalent of "aptitude update" on every run. |
| 21:23 | <@jerith> | Well, whenever it thinks the cache is stale. |
| 21:23 | <@ToxicFrog> | Well, the man page says it should work - packages can be specified as "name" (latest version), or "name-ver" for a specific version |
| 21:24 | <@jerith> | Which is invariably not when I've just updated my repo. |
| 21:24 | <@McMartin> | It sounds like what "should" be done there is yum -C list (pkg) and then yum -C install (pkg.version.number) |
| 21:24 | <@jerith> | But anyway. A new version of foo (which is small) wants to pull in a new version of bar (which is huge). |
| 21:25 | <@jerith> | If I say -C, it won't download the rpm. |
| 21:25 | <@ToxicFrog> | pkg-version; . is for pkg.arch |
| 21:25 | <@jerith> | At least, not last time I tried. |
| 21:25 | <@McMartin> | Well, OK, yum -C install wouldn't work. But with the version marker it should |
| 21:26 | <@McMartin> | (I've had to do this with arch stuff when x86_64 and i386 stuff gets confused post-distro-change) |
| 21:26 | <@ToxicFrog> | It *should* work, according to man yum - "work from cache; do not download or update unless necessary" |
| 21:26 | <@McMartin> | (But the 7->8 hop didn't have that problem, so maybe it was just a 5-7 thing) |
| 21:26 | <@jerith> | I haven't tried it in several months, so maybe it does with new yum. |
| 21:27 | <@jerith> | I don't trust its idea of "necessary", though. |
| 21:28 | <@jerith> | And I would much rather have "install foo" decoupled from "update stuff". |
| 21:28 | <@ToxicFrog> | It is, you just don't like the way it does it |
| 21:29 | <@ToxicFrog> | ...and it's broken, since according to my test which just finished -C still doesn't work as per the man page |
| 21:29 | <@jerith> | It's only decoupled if I go to effort to decouple it, and even then it (historically) does it badly. |
| 21:29 | <@McMartin> | That's a bug, not a design flaw, though. |
| 21:29 | <@McMartin> | Most of my complaints about emerge are actual design flaws |
| 21:29 | <@ToxicFrog> | Yeah. |
| 21:29 | <@McMartin> | And with apt are the legacy of a decade of crappy repo management |
| 21:30 | <@ToxicFrog> | And I'm still not sure what your complaint about the fact that it updates the repos each time needed |
| 21:30 | <@McMartin> | He's in .za, under draconian bandwidth limits. |
| 21:30 | <@ToxicFrog> | Since if you have little enough bandwidth available that this is an issue, you'll have serious trouble actually installing anything in the first place |
| 21:30 | <@jerith> | It's not that bandwidth is limited, it's that it's expensive. |
| 21:30 | <@McMartin> | That's a limit. |
| 21:31 | <@jerith> | If I'm at home, I get 500mb/month and anything over that is about US$0.20/mb. |
| 21:31 | <@McMartin> | That's a comically sharp limit. |
| 21:31 | <@jerith> | At the office, it's somewhat cheaper, so I can do updates there as required. |
| 21:32 | <@jerith> | Even in the package, it's only about half the out-of-bundle cost. |
| 21:32 | <@McMartin> | That's, in fact, "you might as well go to the store and buy a CD every six months" sharp. |
| 21:32 | <@jerith> | My point is that the yum philosophy makes sence when bandwidth is free and unlimited. |
| 21:32 | <@McMartin> | Yes. |
| 21:32 | <@jerith> | *sense |
| 21:33 | <@McMartin> | But, um |
| 21:33 | <@jerith> | But back when yum was new, bandwidth was not free and unlimited even in the US. |
| 21:33 | <@McMartin> | It's upstream that's limited here. |
| 21:33 | <@McMartin> | They're paying for bandwidth. The downstream is only paying for access. |
| 21:34 | <@ToxicFrog> | He means for the users |
| 21:34 | <@McMartin> | That's certainly preferable, in my book, to paying $400US for an install DVD. |
| 21:34 | <@ToxicFrog> | Where downstream is limited |
| 21:34 | <@jerith> | That model is broken, but subtly so. |
| 21:34 | <@ToxicFrog> | ...and downstream is limited in the US too, at least under some providers |
| 21:34 | <@McMartin> | I've had internet access of some kind or another since 1995 and have never run into such things. |
| 21:34 | <@jerith> | Where do you pay $400 for an install DVD? |
| 21:34 | <@ToxicFrog> | Same in canada |
| 21:34 | <@McMartin> | jerith: You do. |
| 21:34 | <@ToxicFrog> | Although the limit is still tens or hundreds of gigabytes |
| 21:35 | <@McMartin> | An install DVD full of stuff is 2.5GB. |
| 21:35 | <@jerith> | Oh, the download. |
| 21:35 | <@ToxicFrog> | jerith: that's what it would cost you to download an install DVD. |
| 21:35 | <@jerith> | In '95 most people were still on dialup. |
| 21:35 | <@McMartin> | I feel quite confident in saying that charging $400 for that is Fucked Up (tm). |
| 21:35 | <@jerith> | Oh, right. You people have ridiculously cheap (by our standards) local phone rates, too. |
| 21:36 | <@McMartin> | It's presumably largely covered by higher flat subscription rates. |
| 21:36 | <@McMartin> | But, uh, yeah, I'll take it. |
| 21:36 | <@jerith> | If I could convice our much-beloved telecomms monopoly to give me a line, I'd be paying about US$10/gb. |
| 21:37 | <@jerith> | A standard phone line here is about $25 or so per month. |
| 21:37 | <@McMartin> | I'm probably paying more than that once you factor in the flat feez. |
| 21:37 | <@McMartin> | It's about $40 here. |
| 21:37 | | * jerith considers. |
| 21:37 | <@McMartin> | (And is in fact the lion's share of my Cost Of Internet) |
| 21:37 | <@jerith> | I think I'm screwing up the exchange rate in my head here... |
| 21:38 | <@McMartin> | Possibly not; that $40 does include unlimited local. |
| 21:38 | <@McMartin> | And it's not like the telcos are going to charge prices that lose them money. |
| 21:38 | <@jerith> | Call it R7.5/$ |
| 21:41 | <@jerith> | Hmm. It's R112/month for copper and basic phone service. |
| 21:41 | <@jerith> | !dice 112/7.5 |
| 21:41 | < DiceBot> | [jerith] 112/7.5 = 14.933333. |
| 21:41 | <@ToxicFrog> | What's the actual currency name? |
| 21:41 | <@jerith> | So $15 |
| 21:41 | <@jerith> | Rand. |
| 21:42 | <@McMartin> | Yeah. I'm paying $40 for a land line, another $20 on top of that for DSL through it, and then about $30 for cellular (which includes long-distance). |
| 21:42 | <@McMartin> | But these numbers never, ever change. |
| 21:43 | <@jerith> | ADSL on top of that (just the plug into the DSLAM, no ISP, no data, no customer-side kit) is a further R152 |
| 21:43 | <@jerith> | !dice 264/7.5 |
| 21:43 | < DiceBot> | [jerith] 264/7.5 = 35.2. |
| 21:43 | <@jerith> | Call it US$35 for the copper. |
| 21:44 | <@McMartin> | I'm not sure what you mean by "copper" here. |
| 21:44 | <@McMartin> | Other than "wiring" |
| 21:44 | < Shoukanjuu> | What are we talking about? |
| 21:44 | <@McMartin> | TCO for Intertrons across nations. |
| 21:44 | <@jerith> | Then it's usually another $20 or so to the ISP to put data through it, at a basic package that gives you 3gb/mo. |
| 21:44 | | * Shoukanjuu pays over 100 dollars for DSL and phone, but they are separate lines, so |
| 21:44 | <@McMartin> | Well, TCS |
| 21:45 | <@McMartin> | Shou: Likewise, if you count the cell as a "separate line" |
| 21:45 | < Shoukanjuu> | I have 2 lines, one of which I am pretty sure runs with the DSL |
| 21:45 | <@jerith> | McMartin: The wire plus the exchange equipment. No services. |
| 21:45 | <@McMartin> | Aha. |
| 21:45 | < Shoukanjuu> | So the DSL line+line 1, plus the second "line" that is different |
| 21:45 | <@McMartin> | Yeah, since I live in an apartment, that cost is effectively part of rent. |
| 21:45 | <@McMartin> | The building is already wired for cable and phone. |
| 21:46 | <@McMartin> | Shou: with modern wiring that shouldn't be necessary; I'm just using a splitter with a filter on it |
| 21:46 | <@jerith> | Well, that's the monthly "line rental" fee. |
| 21:46 | <@McMartin> | Ah. Hm. |
| 21:46 | <@jerith> | You cannot buy a package without a voice line. |
| 21:46 | < Shoukanjuu> | I would imagine |
| 21:46 | <@McMartin> | Well, yeah |
| 21:46 | <@McMartin> | That may be part of my $40. |
| 21:46 | < Shoukanjuu> | But the DSL line comes in at a different place than the phone line does |
| 21:46 | <@jerith> | I would imagine it is. |
| 21:47 | < Shoukanjuu> | I've seen it come a few different ways, in fact |
| 21:47 | <@McMartin> | But (unlike my parents) I didn't have to pay to actually put the wires in. |
| 21:47 | <@jerith> | Oh, and that's the slowest DSL the offer, which is 192kbps. |
| 21:47 | <@jerith> | (Down. Much less up.) |
| 21:47 | < Shoukanjuu> | You poor thing O_o; |
| 21:47 | <@jerith> | Installation is separate and also expensive. |
| 21:47 | <@McMartin> | What kind of service you can get in the States is massively dependent on where you are. |
| 21:48 | <@jerith> | Yeah. |
| 21:48 | <@McMartin> | They built new routers at one point and essentially forced an upgrade on me. |
| 21:48 | <@ToxicFrog> | In vaguely related news, the latest XKCD is hilarious |
| 21:48 | <@McMartin> | "If you don't upgrade the package we're going to charge you *EVEN MORE* because we're going to have to throttle you" |
| 21:48 | <@McMartin> | "And that's work" |
| 21:48 | < Shoukanjuu> | Down here in central florida, 768kbps/126kbps is the lowest |
| 21:48 | <@jerith> | Also, add a 6-30 week wait to have a line installed. |
| 21:48 | < Shoukanjuu> | lol |
| 21:48 | <@jerith> | That's if you're in an area they're willing to service. |
| 21:49 | <@McMartin> | Yay monopolies~ |
| 21:49 | < Shoukanjuu> | mm-hmm |
| 21:49 | < Shoukanjuu> | Now, the cable provider says they were able to provide cable down here |
| 21:49 | < Shoukanjuu> | But the nodes belong to a defunct company |
| 21:49 | <@McMartin> | Between that and the kinds of sploits the ISPs want to violate net neutrality to unleash, it really does seem like "massively regulated competition with the last mile either socialized or a natural monopoly", much like power and water, seems to be the best way to handle comm infrastructure too |
| 21:50 | < Shoukanjuu> | Not to mention we're the only ones out here who want anything higher than 768k |
| 21:50 | < Shoukanjuu> | Mm-hmm |
| 21:52 | <@jerith> | McMartin: Anyways, once you have the line rental, local voice calls are about US$0.06/min, with a minimum charge of a minute and a half. |
| 21:52 | < Shoukanjuu> | Comcast is real bad with that; Embarq not so much, I haven't noticed any throttiling due to P2P here either |
| 21:53 | <@McMartin> | That's our "zone 3" rate, which is generally for calling the next city or two over. |
| 21:53 | <@McMartin> | But still in the same metro area |
| 21:53 | < Shoukanjuu> | Many plans offer "unlimited local calls" |
| 21:53 | <@McMartin> | jerith is in South Africa. |
| 21:53 | <@McMartin> | IIRC |
| 21:53 | <@jerith> | That's within 50km. Outside that is almost double. |
| 21:53 | < Shoukanjuu> | Which I would assume means they dont' count the local minutes, and just offer a flat fee... |
| 21:53 | < Shoukanjuu> | Yeah, that sucks :< |
| 21:55 | <@jerith> | So, to bring my $15 up to your $40, I need to make 400 minutes of calls a month. I think. |
| 21:56 | <@jerith> | Half that if I'm calling another city. |
| 21:56 | <@McMartin> | Yeah. |
| 21:56 | <@McMartin> | Treat it as merely a prereq for DSL, though, and I only have to download a gig in a month to be saving. |
| 21:56 | <@jerith> | How much do you pay your ISP? |
| 21:57 | <@McMartin> | $25. ~400kB down, ~40kB up. |
| 21:57 | <@McMartin> | Goes through the $40 land line. |
| 21:57 | <@jerith> | So you have a base fee of $65/month for interwebs. |
| 21:57 | <@McMartin> | (And is the reason I have it; otherwise I'd just use the cell for everything) |
| 21:58 | < Shoukanjuu> | I think it's somewhere around that for mine |
| 21:58 | < Shoukanjuu> | Phone wasn't required |
| 21:58 | < Shoukanjuu> | with only one line, it's somewhere around 65 a month... |
| 21:59 | <@jerith> | !dice 200/7.5 |
| 21:59 | < DiceBot> | [jerith] 200/7.5 = 26.666667. |
| 21:59 | <@jerith> | I have a base fee of $60, which includes 1gb of data. |
| 21:59 | <@jerith> | Using our best local ISP. |
| 22:00 | <@McMartin> | Didn't you say 500 before? |
| 22:00 | <@McMartin> | 500 isn't enough to let me do a file-release on every project I work on. =/ |
| 22:00 | <@jerith> | My connection is 3g. |
| 22:00 | <@jerith> | Which is worse in every way, except for mobility and the lack of fixed-line requirement. |
| 22:01 | < Shoukanjuu> | 3mbps :O |
| 22:01 | <@jerith> | !dice 450/7.5 |
| 22:01 | < DiceBot> | [jerith] 450/7.5 = 60. |
| 22:01 | <@jerith> | So I pay $60/month for 500mb. |
| 22:02 | <@McMartin> | And that's absolutely comical highway robbery, as we computed before. |
| 22:02 | <@jerith> | !dice 100/7.5 |
| 22:02 | < DiceBot> | [jerith] 100/7.5 = 13.333333. |
| 22:02 | <@jerith> | That much is the monthly installments for my router. |
| 22:03 | <@McMartin> | ...installments? |
| 22:03 | <@McMartin> | I just went down to the electronics store and bought a router for like $40. |
| 22:03 | <@jerith> | The contract includes a router, for which I pay R100/month for the two years. |
| 22:03 | <@jerith> | This is a linksys WRT54G3G, which you can plug a 3g card into. |
| 22:04 | <@McMartin> | Right |
| 22:04 | <@McMartin> | I have a WRT54GL. |
| 22:04 | <@jerith> | At the time (before usb 3g modems where available) it was the only way to get my Linux box online. |
| 22:04 | <@McMartin> | Which I get to own and paied for once, for maybe 3 months' rent. |
| 22:04 | <@McMartin> | paid, even |
| 22:04 | <@jerith> | Also at the time, I could buy said router from a shop I don't like (they're dishonest) for R2400 cash. |
| 22:05 | <@McMartin> | ... OK then. |
| 22:05 | <@jerith> | So I took the R100 over 24 months option. :-) |
| 22:05 | <@McMartin> | Yeah, that's rational. |
| 22:05 | <@McMartin> | But wtf. |
| 22:05 | <@McMartin> | Was there some huge crash in cost since then? |
| 22:05 | <@McMartin> | Or is shipping really that hard? |
| 22:05 | <@McMartin> | Hell, you're closer to China than we are. |
| 22:05 | <@jerith> | A bit, but not so much with that model. |
| 22:06 | <@jerith> | It's a GL with a different case, vodafone branding, a PCMCIA slot for the data card and drivers for said card. |
| 22:06 | <@McMartin> | Right then. |
| 22:06 | <@McMartin> | OK, complaint rescinded |
| 22:06 | <@jerith> | The cost of the router isn't unreasonable. |
| 22:06 | | * McMartin just did a Google store search for what those run here. |
| 22:07 | <@jerith> | What did you get? |
| 22:07 | <@McMartin> | Lowest price I found was $151. |
| 22:07 | <@McMartin> | Which is nearly 4 times what I paid for the WRT54GL. |
| 22:07 | <@jerith> | I'd imagine they're low-volume and the work to get the PCMCIA and drivers in there can't have been insubstantial. |
| 22:08 | <@McMartin> | ($55 online, but $40 in-store) |
| 22:08 | <@jerith> | My GL cost me R550, iirc. |
| 22:08 | <@McMartin> | That's not out of line. |
| 22:08 | | * jerith nods. |
| 22:08 | <@jerith> | Also quite reasonable. |
| 22:08 | <@McMartin> | There's a huge price range on them. |
| 22:08 | <@McMartin> | It seems to go from $40 to $150. |
| 22:08 | <@jerith> | That R2500 was also nearly two years ago. |
| 22:08 | <@McMartin> | But anyone paying $150 for a GL needs their head examined. |
| 22:09 | <@jerith> | Hmm, first Google hit for the G3G locally is R2235. |
| 22:10 | <@jerith> | The next one that seems to sell them is R2100. |
| 22:10 | <@jerith> | So they haven't dropped /that/ much in price. |
| 22:10 | <@jerith> | But anyway. |
| 22:10 | <@jerith> | Let's look at the value of the currency. |
| 22:11 | <@jerith> | When I look at tech job salaries and basic foodstuffs, the exchange rate shoulod be much closer to R3-4/$. |
| 22:11 | <@jerith> | So in real terms, double everything I just said. |
| 22:23 | < Shoukanjuu> | Hrm. |
| 22:24 | < Shoukanjuu> | So I want to have a pdf download when you click a link. |
| 22:25 | < gnolam> | What's the problem? |
| 22:25 | <@C_tiger> | Shoukanjuu what browser? |
| 22:25 | < Shoukanjuu> | I don't know how |
| 22:26 | < Shoukanjuu> | The majority of people are still most likely using IE |
| 22:26 | <@McMartin> | A simple <a href="whatever.pdf">document here</a> should do it. |
| 22:26 | <@McMartin> | Maybe with instructions to right-click and choose Save As. |
| 22:26 | <@C_tiger> | Yeah, instructions are probably best then. |
| 22:26 | < Shoukanjuu> | And where to put the file .-. |
| 22:27 | < Shoukanjuu> | I'm barely a beginner in this |
| 22:27 | < Shoukanjuu> | Definitely not meant to be a publisher this quick |
| 22:27 | <@McMartin> | So, do you have links between web pages down yet? |
| 22:27 | <@McMartin> | You basically treat the PDF as if it were Just Another Web Page Or Link Target. |
| 22:27 | < Shoukanjuu> | Adn as long as the file is in the directory |
| 22:28 | <@McMartin> | Yeah. In some browsers it will open in-place by default, in others it will download |
| 22:28 | < Shoukanjuu> | That you say it is in |
| 22:28 | < Shoukanjuu> | Okay |
| 22:28 | <@McMartin> | But IMO that's for the user to decide, not you. |
| 22:28 | < Shoukanjuu> | Thank you |
| 22:28 | < Shoukanjuu> | Click to view, right-click, save as... to save >_> |
| 22:28 | < Shoukanjuu> | But that wouldn't be all inclusive |
| 22:28 | <@C_tiger> | I think there's a javascript trick to force saves... but I forget. |
| 22:28 | < Shoukanjuu> | Because like you said ^ |
| 22:29 | < Shoukanjuu> | And I don't know savascript >.> |
| 22:29 | < Shoukanjuu> | java* |
| 22:29 | < Shoukanjuu> | Because I was told to learn python first <.< |
| 22:29 | < Shoukanjuu> | Which I haven't gotten around to brushing up on and continuing x_X |
| 22:29 | <@McMartin> | The tricks have nothing to do with the langauge |
| 22:29 | <@McMartin> | And more to do with stuff that will break for 40% of your viewers. |
| 22:30 | | * McMartin is firmly in the "This is the user's decision, not yours" bin. |
| 22:30 | <@C_tiger> | Yeah, it's not a good idea, Shoukanjuu but I'm just saying it's there. |
| 22:44 | < Shoukanjuu> | I'm just doing the instruction thing |
| 22:44 | < Shoukanjuu> | "Right-Click, 'Save as...' |
| 22:53 | | Steve is now known as AnnoDomini |
| 22:56 | | gnolam [lenin@Nightstar-10613.8.5.253.static.se.wasadata.net] has quit [Ping Timeout] |
| 23:01 | | gnolam [lenin@Nightstar-10613.8.5.253.static.se.wasadata.net] has joined #Code |
| 23:01 | | mode/#code [+o gnolam] by ChanServ |
| 23:07 | < Shoukanjuu> | Works :> |
| 23:07 | < Shoukanjuu> | Aaaand let's upload the files to the wrong folder because they're the same length, sure |
| 23:07 | < Shoukanjuu> | That'll work... *kicks himself* |
| 23:17 | | AnnoDomini [AnnoDomini@Nightstar-29227.neoplus.adsl.tpnet.pl] has quit [Quit: "I am Magellan-R-MGE. I see your group is missing a Hygiene officer."] |
| 23:43 | <@McMartin> | Shou: While we're at it, that's also how you set up any other kind of special file for download. |
| 23:44 | <@McMartin> | It's how I publish my adventure games and source packages and such. |
| 23:51 | <@McMartin> | http://blog.wired.com/monkeybites/2008/04/microsoft-datab.html |
| 23:54 | | Vornicus-Latens is now known as Vornicus |
| 23:57 | < Shoukanjuu> | that works, cool |
| --- Log closed Tue Apr 29 00:00:36 2008 |