| --- Log opened Fri May 17 00:00:29 2013 |
| 00:01 | | Derakon [chriswei@Nightstar-a3b183ae.ca.comcast.net] has quit [[NS] Quit: leaving] |
| 00:07 | | VirusJTG_ [VirusJTG@Nightstar-09c31e7a.sta.comporium.net] has joined #code |
| 00:08 | | VirusJTG [VirusJTG@Nightstar-09c31e7a.sta.comporium.net] has quit [Ping timeout: 121 seconds] |
| 00:22 | | You're now known as TheWatcher[T-2] |
| 00:29 | | You're now known as TheWatcher[zZzZ] |
| 00:30 | <~Vornicus> | it bogs down a lot for the last planets in a system though. |
| 00:31 | | himi [fow035@D741F1.243F35.CADC30.81D435] has joined #code |
| 00:31 | | mode/#code [+o himi] by ChanServ |
| 00:48 | | Derakon[AFK] is now known as Derakon |
| 00:51 | | ErikMesoy is now known as ErikMesoy|sleep |
| 00:54 | | cpux [cpux@Nightstar-98762b0f.dyn.optonline.net] has joined #code |
| 00:54 | | mode/#code [+o cpux] by ChanServ |
| 00:55 | <&Derakon> | What are you doing? |
| 01:38 | <~Vornicus> | generating the production possibilities frontier for a space 4x empire. A lot of good stuff happens inside. Most importantly I'm taking the minkowski sum of the various planets' ppfs and removing things that aren't "on" the frontier. |
| 01:42 | <~Vornicus> | so near the end of a typical run, I'll have a planet with 300 elements in its ppf and the rest of the system has 50-100k elements and I have to merge every pair. |
| 01:42 | <&Derakon> | That sounds like a lot of "try all combinations" kinds of things. |
| 01:42 | <~Vornicus> | Yep. |
| 01:44 | <~Vornicus> | There's a lot of stuff to try to combine. I suspect a big part of the memory difficulty at this point is that there's so much to do. |
| 01:44 | <&Derakon> | What all are you storing? |
| 01:50 | <~Vornicus> | All told the storage is all the ppf nodes. These store their values (a 3-tuple of integers), and the decisions made to reach those values (each planet has a name, a number for how many system-wide improvement facilities on that planet, and an up-to-8-character string describing the facility loadout on that planet) |
| 01:51 | <~Vornicus> | During the system buildup phase, which I'm currently working on, there are 28 categories that these ppf nodes fall into. At the end, 20 of those categories are invalidated (because they have been counting on a certain number of system-wide improvement facilities, and that number hasn't been met), and the remaining 8 are merged together and the frontier is simplified. |
| 01:56 | <~Vornicus> | There are multiple other simplification steps involved during the buildup process; every time I add two frontiers (to produce, say, the production possibilities of two planets together), or merge two frontiers (because, say, there's one system-wide facility between the two planets and the two frontiers represent whether it's on planet A or B), I simplify it, because usually there aren't nearly |
| 01:56 | <~Vornicus> | as many things /on/ the frontier as there are /within/ it. |
| 02:02 | <~Vornicus> | The frontier gets pretty big - 50-120k points in each category, with all the stuff listed above. |
| 02:04 | | ToxicFrog [ToxicFrog@ServerAdministrator.Nightstar.Net] has quit [[NS] Quit: ZNC - http://znc.in] |
| 02:13 | <~Vornicus> | and generating the frontier with one additional planet involves another minkowski sum, so here I am, figuring that maybe if I could make my code a little less reliant on nested adjustable-size containers I could rewrite it in C/C++ and have something more performant. |
| 02:13 | <&Derakon> | What's the goal of doing all this anyway? |
| 02:17 | <~Vornicus> | I always have a hard time deciding where to put things in space 4x games;by generating all the "best" layouts I can then use that data to choose the one that best suits my goals and build to that. |
| 02:17 | <@Reiv> | hn, I'd have gone for genetic algorathms there |
| 02:18 | <@Reiv> | You may not get the absolutely mathematically optimal solution, but you'll still get a Pretty Good one. |
| 02:18 | <~Vornicus> | if I knew how to do those at all I would be! |
| 02:20 | <@Reiv> | Take a bunch of random assortments. Work out which ones work best. From a subset of the best, randomly splice two sets together at a time a bunch of times each to get a new set of assortments. Loop. |
| 02:21 | <&Derakon> | Genetic algorithm: you have a set of inputs; generate a bunch of random values for the inputs; compute a cost function for each set of values ("how good are these values?"). Keep the best ones, randomly permute them, repeat. |
| 02:22 | <&Derakon> | Honestly the main thing here is making the cost function. |
| 02:22 | <&Derakon> | If you have that, you can also just plug it into Simplex. |
| 02:22 | <&Derakon> | (There's a Simplex implementation in scipy) |
| 02:24 | <~Vornicus> | Well, cost function, and figuring out how to randomize this stuff. |
| 02:24 | <&Derakon> | For Simplex all you need is the cost function and a starting guess. |
| 02:31 | | * Vornicus thinks at it |
| 02:32 | <~Vornicus> | okay looking around I see scipy.optimize |
| 02:36 | | Turaiel[Offline] is now known as Turaiel |
| 02:37 | <&Derakon> | Yeah, scipy.optimize.fmin. |
| 02:37 | <&Derakon> | params = scipy.optimize.fmin(costFunc, initialGuess) |
| 02:40 | <~Vornicus> | this cost function is going to be epic. |
| 02:41 | <&Derakon> | The main thing is the number of parameters -- the more there are, the more difficult the optimization becomes. |
| 02:41 | <~Vornicus> | Part of it is there's so many constraints |
| 02:45 | <~Vornicus> | --and ones I don't necessarily know how to explain to the program. How /do/ I tell it that all the inputs have to be integers, that certain groups of values must add up to at most certain other values, that... |
| 02:48 | <&Derakon> | If you set the cost of an invalid combination to infinity, then the optimizer should avoid it. :) |
| 02:48 | <~Vornicus> | heh. |
| 02:48 | <&Derakon> | But for integer-only constraints, mm, you could multiply the inputs by some large number and then cast to int. |
| 02:49 | <&Derakon> | So e.g. if the program asks you for the cost of .00012, you multiply by 10e6 to get 100.2, cast to 100, there's your "actual input". |
| 02:49 | <&Derakon> | I have to do something similar at work since the step size that fmin uses is typically too small to plug directly into my cost function. |
| 03:02 | | * Vornicus gets to work on cost function, and while he's at it makes sure that his install has scipy |
| 03:28 | <@Alek> | darn scipy it does. |
| 03:32 | | * Vornicus patpats alek |
| 03:34 | | Turaiel is now known as Turaiel[Offline] |
| 03:35 | | VirusJTG_ [VirusJTG@Nightstar-09c31e7a.sta.comporium.net] has quit [[NS] Quit: Program Shutting down] |
| 03:38 | | Kindamoody[zZz] is now known as Kindamoody |
| 03:41 | | RichyB [richardb@58734C.5279B7.EA7DF8.107330] has joined #code |
| 03:54 | | Turaiel[Offline] is now known as Turaiel |
| 03:58 | | RichyB [richardb@58734C.5279B7.EA7DF8.107330] has quit [Ping timeout: 121 seconds] |
| 03:59 | | RichyB [richardb@58734C.5279B7.EA7DF8.107330] has joined #code |
| 04:01 | <~Vornicus> | okay. each planet has 3 integers and 3 booleans as cost function input. Each system has three booleans as cost function input. I /believe/ that covers the bases. |
| 04:03 | | RichyB [richardb@58734C.5279B7.EA7DF8.107330] has quit [Connection closed] |
| 04:07 | <@Reiv> | HEY VORN |
| 04:07 | <~Vornicus> | shit |
| 04:07 | <@Reiv> | Do you remember doing the math on 1d6-(diepool of 12d6)? |
| 04:07 | <~Vornicus> | Yes. |
| 04:08 | <@Reiv> | How hard would it be to do 2d6-(diepool of 12d6)x2 ? |
| 04:08 | <~Vornicus> | So doubling the value of the die pool? |
| 04:08 | <~Vornicus> | Not hard. |
| 04:09 | <@Reiv> | And turning the d6 into 2d6 |
| 04:09 | <@Reiv> | This being a 'double everything, but not' |
| 04:09 | <~Vornicus> | oh, I see. Yes, not hard. |
| 04:09 | <@Reiv> | Actually the numbers in question are 9d6, 11d6 and 13d6 but meh |
| 04:10 | <~Vornicus> | I forget, what's the threshold value on the pool? |
| 04:10 | <@Reiv> | This is prodding at an RPG that probably thought it was being clever... but probably wasn't. |
| 04:19 | <@celticminstrel> | What does this diepool stuff mean? |
| 04:20 | <@Reiv> | Die pools are where you roll a bunch of dice and then, rather than adding them together, count the number of dice above a certain threshold. |
| 04:21 | <@Reiv> | Eg, 4+ on a d6 |
| 04:22 | | ToxicFrog [ToxicFrog@ServerAdministrator.Nightstar.Net] has joined #code |
| 04:22 | | mode/#code [+ao ToxicFrog ToxicFrog] by ChanServ |
| 04:22 | <@Reiv> | Gives you a binomial distribution rather than a standard distribution, which is a bit flattened out, and lets you roll lots of dice. |
| 04:25 | <&ToxicFrog> | \o/ |
| 04:25 | <&ToxicFrog> | scan: resilver in progress since Thu May 16 23:10:23 2013 |
| 04:25 | <&ToxicFrog> | 129G scanned out of 3.63T at 151M/s, 6h43m to go |
| 04:25 | <&ToxicFrog> | 31.2G resilvered, 3.48% done |
| 04:27 | <@Alek> | :P |
| 04:50 | <&Derakon> | What are you scanning for? |
| 04:55 | <@celticminstrel> | I see... |
| 05:02 | <&ToxicFrog> | Derakon: it's a RAID resync, basically. |
| 05:05 | <&Derakon> | Ah. |
| 05:10 | <&ToxicFrog> | ("resilvering" is the general term for all ZFS drive integration operations) |
| 05:12 | | Derakon is now known as Derakon[AFK] |
| 05:29 | <@celticminstrel> | ...hm. Finding the perpendicular vectors... surely there's a general way to do this... |
| 05:29 | <~Vornicus> | What do you have that you're trying to find perpendiculars to? |
| 05:30 | <@celticminstrel> | Two dimensions, both components in (-1,0,1). |
| 05:30 | <@celticminstrel> | Obviously no (0,0). >_> |
| 05:31 | <~Vornicus> | The perpendicular vector to (a, b) is (-b, a), or (b, -a) |
| 05:31 | <@celticminstrel> | Excellent! Thanks. |
| 05:36 | | Derakon[AFK] is now known as Derakon |
| 06:15 | | Derakon is now known as Derakon[AFK] |
| 06:23 | | Derakon [Derakon@Nightstar-a3b183ae.ca.comcast.net] has joined #code |
| 06:23 | | mode/#code [+ao Derakon Derakon] by ChanServ |
| 06:25 | | Kindamoody is now known as Kindamoody|out |
| 06:26 | | Derakon[AFK] [Derakon@Nightstar-a3b183ae.ca.comcast.net] has quit [Ping timeout: 121 seconds] |
| 06:27 | | Derakon [Derakon@Nightstar-a3b183ae.ca.comcast.net] has quit [Ping timeout: 121 seconds] |
| 06:27 | | Derakon [Derakon@31356A.8FA1FE.CF2CE9.D6CF77] has joined #code |
| 06:27 | | mode/#code [+ao Derakon Derakon] by ChanServ |
| 06:41 | | * Vornicus blinks, derps. could just, you know, insist every planet specializes, and then there's only like 50 loadouts per planet instead of the 6000 he was using before, |
| 06:47 | <@froztbyte> | emperor vorn |
| 06:47 | <~Vornicus> | that's a me |
| 06:56 | | AverageJoe [evil1@Nightstar-4b668a07.ph.cox.net] has joined #code |
| 07:00 | | Turaiel is now known as Turaiel[Offline] |
| 07:20 | | celticminstrel [celticminst@Nightstar-e83b3651.cable.rogers.com] has quit [[NS] Quit: KABOOM! It seems that I have exploded. Please wait while I reinstall the universe.] |
| 07:24 | | himi [fow035@D741F1.243F35.CADC30.81D435] has quit [Ping timeout: 121 seconds] |
| 07:25 | | * Vornicus gets it to do a whole system in about a minute. Now comes the ridiculous part: the specialize thing hasn't reduced the number of items in the final result list terribly much. |
| 07:27 | <~Vornicus> | THis is probably because, well, specialization /works/. I suspect that in the unspecialized version, your typical "optimal" loadout will only have one or two non-specialized planets. |
| 07:30 | <~Vornicus> | So this is overall Bad News, because it means my next step is a lot grumpier. right about now more than 8 cores would be nice~ |
| 07:37 | <@froztbyte> | crowd computing!!~ |
| 07:37 | <~Vornicus> | heh |
| 07:41 | <~Vornicus> | (though of course I'd need to actually engineer that, and I've never done high power stuff like that) |
| 07:57 | | ErikMesoy|sleep is now known as ErikMesoy |
| 08:15 | <~Vornicus> | oh shit, no, I can't parallelize that, can I, it's a thing that demands order. |
| 08:22 | <@Azash> | Depends on whether it's enforcable order ("these blocks must be in order") or execution must be in order |
| 08:22 | <@Azash> | In the former case, you can parallelize computation as long as you have a way to order the results |
| 08:22 | <@Azash> | Disclaimer: I haven't seen what the problem in question is |
| 08:28 | <~Vornicus> | The particular part of the problem that requires order, requires execution order; I sort the list and then filter it based on previous results. Technically I can make an n^2 algorithm instead of the existing n ln n but I'm already hitting billions of elements, so... |
| 08:39 | <@froztbyte> | are you generating stuff from that pipeline, with each previous one influencing the next one? |
| 08:40 | <~Vornicus> | All previous ones, influencing the filter status of the next one. Well, some of the previous ones. |
| 08:40 | <@froztbyte> | mmmmm |
| 08:41 | <~Vornicus> | (I have a list of 3-vectors; I'm filtering out items that are dominated by something: an item x is dominated by another item d if for all i, x_i <= d_i) |
| 08:43 | <~Vornicus> | (this works best if you sort by (x_0, x_1 + x_2), descending; you can maintain a pair of lists for x_1 and x_2 and bisect to determine if a thing is dominated and what things are no longer necessary to check because another item will dominate all the ones that the ones you're deleting will.) |
| 08:53 | <~Vornicus> | So every item in the list is either dominated or changes the 'beach' line that will decide if future items are dominated. |
| 08:53 | <@froztbyte> | haha |
| 08:53 | <@froztbyte> | yeah |
| 08:53 | <@froztbyte> | that's a bit of a hard problem to feed out quickly |
| 09:03 | <~Vornicus> | So parallelizing is unfortunately not on the menu for the hardest task. I am starting to wonder why there's so many undominated items at the system level - it seems like there should be fewer, but I can't see from here how. |
| 09:03 | | AverageJoe [evil1@Nightstar-4b668a07.ph.cox.net] has quit [[NS] Quit: Leaving] |
| 09:07 | <@froztbyte> | well, I'm entirely unfamiiar with your code, even more so general algorithmic generation stuff because I haven't done anything like it in a loooooooong time |
| 09:07 | <@froztbyte> | so no helpful ideas from me :) |
| 09:11 | | You're now known as TheWatcher |
| 09:50 | | himi [fow035@Nightstar-5d05bada.internode.on.net] has joined #code |
| 09:50 | | mode/#code [+o himi] by ChanServ |
| 09:55 | <@froztbyte> | http://linkxs.org/freezing-a-raspberry-pi-in-liquid-nitrogen/ |
| 10:21 | <@TheWatcher> | .... |
| 10:21 | <@TheWatcher> | neat! |
| 10:40 | <@froztbyte> | I wonder if it's the ethernet controller that goes |
| 10:40 | <@froztbyte> | or the pins that stop making contact |
| 10:46 | < Syk> | maybe the USB bus freezing |
| 10:46 | < Syk> | :P |
| 10:46 | <@froztbyte> | no |
| 10:46 | <@froztbyte> | that controller isn't on the USB bus, is it? |
| 10:46 | < Syk> | yes it is |
| 10:47 | <@froztbyte> | heh |
| 10:47 | <@froztbyte> | interesting |
| 10:47 | < Syk> | that's why ethernet is shit on the raspi |
| 10:47 | <@froztbyte> | not that I've had chance to play much with that pi I have |
| 10:47 | < Syk> | froztbyte: http://stream1.gifsoup.com/view4/1805762/seattle-snow-bus-o.gif |
| 10:47 | < Syk> | froztbyte: illustration of the USB bus at -97C |
| 10:47 | <@froztbyte> | lawl |
| 10:47 | <@Tamber> | bus bus! |
| 10:47 | <@froztbyte> | I should link you to tkap's crashcam next time it's winter in america |
| 10:49 | <~Vornicus> | "At -105C the watch command for temperature got a segfault" |
| 10:51 | <@TheWatcher> | argh, manually maintaining refcounts in metadata trees: not fun |
| 10:52 | <~Vornicus> | Comments: "5GHz! Yeah, let's run Crysis on it!" |
| 10:52 | <@froztbyte> | lawl. |
| 10:52 | <@froztbyte> | TheWatcher: read-lock-write-unlock-panic loops? |
| 10:53 | <@TheWatcher> | Nah "shit, I missed another reference from there" problems |
| 10:53 | <@froztbyte> | hehe |
| 10:54 | <@froztbyte> | I really wanna adopt this song for when I run into programming/systems problems: http://www.youtube.com/watch?v=xlxrIq2dtI0 |
| 10:54 | <@froztbyte> | it's just not quite directly usable |
| 10:54 | <@TheWatcher> | (unfortunately, don't have time to write the interface for managing these things properly yet. I may regret that) |
| 10:54 | <@froztbyte> | Syk: that gif finally finished loading |
| 10:55 | <@froztbyte> | Syk: that's certainly one way to get ice/snow off the front of the bus. |
| 11:21 | | Vorntastic [Vorn@Nightstar-221158c7.sd.cox.net] has joined #code |
| 11:53 | | Vorntastic [Vorn@Nightstar-221158c7.sd.cox.net] has quit [Client closed the connection] |
| 13:11 | | * TheWatcher eyes this |
| 13:12 | <@TheWatcher> | @{$capabilities}{keys %{$rolecaps}} = values %{$rolecaps}; |
| 13:12 | | Vornicus [vorn@31356A.68201E.EB0611.E0094F] has quit [[NS] Quit: ] |
| 13:13 | <@TheWatcher> | That somehow feels a lot nastier than it should. |
| 13:34 | <@froztbyte> | perrrrrrrls |
| 13:34 | <@froztbyte> | so of course it is ;D |
| 13:53 | <@TheWatcher> | :P |
| 14:52 | <@froztbyte> | http://sushee.no-ip.org/opensourceisnotawarzone.txt |
| 14:52 | <@froztbyte> | TheWatcher: ^ to perl's credit |
| 14:58 | | * McMartin also laughs at XCOM |
| 14:58 | <&McMartin> | I can't even tell if this is a bug |
| 14:59 | | * McMartin detonated some explosive scenery by doing suppression-fire against an enemy in front of it |
| 15:00 | | ToxicFrog is now known as ToxicFrog|W`rkn |
| 15:00 | <&ToxicFrog|W`rkn> | Orias is looking a bit unwell :/ |
| 15:00 | <&ToxicFrog|W`rkn> | The resilver is still running, it's detected read errors on one drive, and /dev/sdd has vanished entirely. |
| 15:00 | < ErikMesoy> | McMartin: Doesn't sound like a bug to me. Scenery is explosive, missed shots can hit scenery. |
| 15:00 | <&ToxicFrog|W`rkn> | Fortunately it wasn't being used for anything. |
| 15:01 | <&ToxicFrog|W`rkn> | Which makes me wonder even more why it vanished. I wonder if the cats unplugged it or something. |
| 15:05 | <@froztbyte> | hehe |
| 15:11 | <&ToxicFrog|W`rkn> | (on the plus side the only casualties - so far - are some PS2 game images which can easily be replaced by re-ripping the discs) |
| 15:56 | <&ToxicFrog|W`rkn> | <3 zfs |
| 15:56 | <@froztbyte> | haha |
| 15:56 | <@froztbyte> | yeah. |
| 15:56 | <@froztbyte> | I'm so glad we're getting better filesystems these days |
| 15:58 | <&ToxicFrog|W`rkn> | Yeah, on extN this would - if it got reported at all - be "some of your data is missing. Good luck finding out which." Assuming mdadm itself didn't freak out. |
| 15:59 | < Xon> | ToxicFrog|W`rkn, wait, it would report it? |
| 15:59 | <@froztbyte> | ToxicFrog|W`rkn: well, this copy is *still* going |
| 16:00 | <@froztbyte> | the one from the previous host |
| 16:00 | <&ToxicFrog|W`rkn> | scan: resilver in progress since Thu May 16 23:10:23 2013 |
| 16:00 | <&ToxicFrog|W`rkn> | 3.58T scanned out of 3.63T at 88.3M/s, 0h9m to go |
| 16:00 | <&ToxicFrog|W`rkn> | 872G resilvered, 98.71% done |
| 16:00 | <&ToxicFrog|W`rkn> | (that 9m is a lie, it's been saying that for the past three hours) |
| 16:01 | < Xon> | I'ld love for something like ZFS for windows for my VM host. Not this bullshit ReFS which has more limitations than you can poke and massive missing features with it's intergration with Storage spaces (aka volume virtualisation) |
| 16:03 | < Xon> | McMartin, new or old X-COM? In old X-COM if there is anything still standing you wheren't using enough gun/explosives |
| 16:10 | < ErikMesoy> | X-COM is old. The new one is called XCOM. |
| 16:11 | < ErikMesoy> | And speaking of which, I just caught a unit teleporting in XCOM. Was entering the Battleship, cleared starting path to where it splits, sent a rocket-grenade-siege dude half a screen left to blast anything coming that way, had the rest of the team go right. Something came up behind Captain Firepower. |
| 16:11 | < ErikMesoy> | There was no path it could have taken to get there from the unscouted parts of the ship. |
| 16:12 | <&ToxicFrog|W`rkn> | Xon: why are you hosting VMs on windows? |
| 16:13 | <&ToxicFrog|W`rkn> | (also, semi-serious suggestion: ZFS supports exporting ZFS storage as iSCSI targets. I think windows supports iSCSI.) |
| 16:15 | < Xon> | ToxicFrog|W`rkn, I work with windows a lot at work, so home lab |
| 16:17 | <&ToxicFrog|W`rkn> | (...windows VMs in a windows VM host which is itself a VM stored on ZFS) |
| 16:19 | < Xon> | actually been thinking about something like that =p |
| 16:20 | | Syk [the@Nightstar-3bbfddc4.iinet.net.au] has quit [[NS] Quit: leaving] |
| 16:20 | | Syka [the@Nightstar-3bbfddc4.iinet.net.au] has joined #code |
| 16:24 | | Kindamoody|out is now known as Kindamoody |
| 16:28 | <&ToxicFrog|W`rkn> | But yeah, impressions of zfs so far: it owns owns owns. There are a few obvious missing features, but even so, it's so much better than anything else I've used, btrfs included. |
| 16:30 | < Syka> | zfs is quite neat, yeah :3 |
| 16:30 | | celticminstrel [celticminst@Nightstar-e83b3651.cable.rogers.com] has joined #code |
| 16:30 | | mode/#code [+o celticminstrel] by ChanServ |
| 16:30 | < Xon> | ToxicFrog|W`rkn, being better than a file system still requiring massive development work done on it isn't exactly the most challanging feat =p |
| 16:31 | < Syka> | well, zfs isn't exactly 'brand new' |
| 16:31 | <@froztbyte> | zfs is pretty stable, yeah |
| 16:31 | < Syka> | i think it's more that btrfs is kinda crap right now |
| 16:31 | <@froztbyte> | (in most flavours) |
| 16:32 | <@froztbyte> | ToxicFrog|W`rkn: so, a bit of info re btrfs vs zfs |
| 16:32 | <@froztbyte> | the former actually followed the latter by a short while |
| 16:32 | <@froztbyte> | not least because of the license issues around zfs |
| 16:33 | <@froztbyte> | and there's also a couple of different problems which they're taking on |
| 16:33 | <@froztbyte> | but for day-to-day use on a desktop, I've found it reasonably decent so far |
| 16:33 | <@froztbyte> | been rocking btrfs on various systems for around 2 years now |
| 16:33 | <@froztbyte> | likho% mount | grep btr |
| 16:33 | <@froztbyte> | /dev/sda2 on / type btrfs (rw,relatime,ssd,space_cache) |
| 16:34 | <@froztbyte> | vandali% mount | grep btr |
| 16:34 | <@froztbyte> | /dev/disk/by-uuid/79874bec-ddc2-403b-a34b-2076acccdc80 on / type btrfs (rw,relatime,ssd,nospace_cache) |
| 16:34 | <@froztbyte> | etc |
| 16:34 | <&ToxicFrog|W`rkn> | Xon et al: well, yes, my point is more that btrfs seems to be the only fs that is even close to being a contender to zfs at the moment, and of those two, I prefer zfs |
| 16:34 | < Syka> | heh |
| 16:35 | < Syka> | i have "tank on /tank type zfs (rw,noatime)" |
| 16:35 | < Xon> | btrfs was initially promised under a hopelessly niave timeframe =p |
| 16:35 | <&ToxicFrog|W`rkn> | orias:/var/log # zfs list | wc -l |
| 16:35 | <&ToxicFrog|W`rkn> | 19 |
| 16:35 | <&ToxicFrog|W`rkn> | Don't think I'm going to paste all of that in here~ |
| 16:36 | < Syka> | all my zfs tanks are small time, though |
| 16:36 | < Syka> | I think the biggest i've got atm is 2x 2TB |
| 16:36 | < Syka> | <v<; |
| 16:36 | <&ToxicFrog|W`rkn> | But yeah, the only stuff I've run into so far that it feels like it should have but doesn't is raidz reshaping and zfs mv. |
| 16:37 | <&ToxicFrog|W`rkn> | NAME SIZE ALLOC FREE CAP DEDUP HEALTH ALTROOT |
| 16:37 | <&ToxicFrog|W`rkn> | orias 7.25T 3.63T 3.62T 50% 1.00x DEGRADED - |
| 16:38 | <@froztbyte> | ToxicFrog|W`rkn: it kinda sucks that 1) you're not on the storage team, 2) that you'd be under NDA anyway |
| 16:38 | < Syka> | my home one is only 1TB :( |
| 16:38 | <@froztbyte> | I would love to ask some questions of the goog storage people :( |
| 16:38 | < Syka> | froztbyte: haven't Google publicly released a whole load of that |
| 16:38 | <@froztbyte> | Syka: not in any useful form |
| 16:38 | <@froztbyte> | go right ahead to try implement it :) |
| 16:38 | <@froztbyte> | (I recall 2 projects which already tried, and failed somewhat) |
| 16:40 | < Syka> | froztbyte: wait, are you meaning how it's used, or how the servers do it |
| 16:40 | < Syka> | since we already know how the servers do it |
| 16:40 | < Syka> | because Google keep publishing papers on it |
| 16:41 | < Syka> | froztbyte: http://en.wikipedia.org/wiki/Google_File_System |
| 16:41 | | Turaiel[Offline] is now known as Turaiel |
| 16:42 | < Syka> | it seems to explain the design behind it pretty good |
| 16:42 | <@froztbyte> | Syka: now find an implementation based off of that. |
| 16:42 | < Syka> | well, such an implementation would be pointless |
| 16:42 | < Xon> | <Syka> my home one is only 1TB :( |
| 16:42 | < Xon> | ~50TB on my home fileserver/vm host=p |
| 16:42 | <@froztbyte> | pointless to whom? |
| 16:43 | < Syka> | froztbyte: pointless to anyone who's not plugging into Google's stuff |
| 16:43 | <@froztbyte> | why? |
| 16:43 | < Syka> | since GFS is at least... ten years old now, probably |
| 16:43 | < Syka> | you could just use the insane Sequoia system! :D |
| 16:43 | <@froztbyte> | XFS is 20 years old |
| 16:44 | <@froztbyte> | and zfs is 8 years old |
| 16:44 | <@froztbyte> | so age of filesystem has no relevance |
| 16:44 | <@froztbyte> | I'm asking you why you think it'd be pointless |
| 16:45 | < Syka> | because that sort of system has had a lot more development in the past 10 years, than traditional FSs have? |
| 16:45 | <@froztbyte> | have you used any of them? |
| 16:45 | <@froztbyte> | them being gluster, ceph, gfs (the RH one), etc |
| 16:49 | | Syka [the@Nightstar-3bbfddc4.iinet.net.au] has quit [Ping timeout: 121 seconds] |
| 16:50 | < Xon> | cluster-file systems tend to relax a lot of hard problems and can leverage native fs for actual storage =p |
| 16:50 | < Xon> | largely because POSIX locking isn't particularly kind for horizontal scaling =p |
| 16:52 | <@froztbyte> | yeah, there's that too |
| 16:52 | <&ToxicFrog|W`rkn> | Hmm. getmail is nicer than fetchmail in many ways, but also brutally slow. |
| 16:54 | < Xon> | froztbyte, to be fair; virtualised SANs are something which used to require a /lot/ of specialised hardware to work. But x86 generic CPUs have improved enough that they can do the work better and faster |
| 16:54 | <@froztbyte> | Xon: believe me, I know |
| 16:55 | < Xon> | heh |
| 16:55 | <@froztbyte> | Xon: the environment I've worked in over the last 4 years has let me touch on a whole bunch of different things |
| 16:55 | <@froztbyte> | and I'm a quick learner |
| 16:55 | <@froztbyte> | so I now have a ridiculously cool skillset :) |
| 16:55 | < Xon> | nice |
| 16:55 | < Xon> | I've been sitting & spinning in the one job for a bit. will likely start looking for a new job within a year or so |
| 16:56 | <@froztbyte> | finishing up here either next week friday, or the one thereafter |
| 16:56 | <@froztbyte> | which will be determined by whether my incompetent colleagues manage to feel satisfied that they know something after next week's Q&A |
| 16:57 | <@froztbyte> | (which Incompetent Prime still hasn't even acknowledged the meeting invite for) |
| 16:57 | <@froztbyte> | (the manager) |
| 16:58 | <@froztbyte> | without going into much further detail I'll just say that I'm thankful for the last couple of months teaching me many more ways to *not* do things |
| 16:59 | <@froztbyte> | (lack of detail because it'll just sound like a lot of whining due to present bitterness) |
| 17:01 | <&ToxicFrog|W`rkn> | 98.80% done! |
| 17:01 | <&ToxicFrog|W`rkn> | This is still going to be running when I get home, isn't it. |
| 17:02 | <@froztbyte> | only if your cats don't kill the power |
| 17:03 | <&ToxicFrog|W`rkn> | I would say "UPS supremacy", except the cats have on at least one occasion powered off the UPS itself |
| 17:03 | <@froztbyte> | :D |
| 17:03 | <&ToxicFrog|W`rkn> | There is now a protective cover over the controls, but I wouldn't put it past them to find some other way |
| 17:03 | <@froztbyte> | lightning made my UPS a somewhat shitty inline power smoother :( |
| 17:03 | <@froztbyte> | (somewhat shitty because the cap lasts 3s, tops) |
| 17:04 | <@froztbyte> | (the relay is also not that healthy anymore) |
| 17:04 | <@froztbyte> | (and the battery is completely fried) |
| 17:04 | <&ToxicFrog|W`rkn> | They love using the server as a jumping pad to reach the window, too, and some of the internal SATA cables are a bit loose, hence my suspicions about /dev/sdd. |
| 17:05 | <&ToxicFrog|W`rkn> | Can't wait until I can replace the server core, too, this upgrade just replaced the disks. |
| 17:05 | <@froztbyte> | hehe |
| 17:05 | <@froztbyte> | yeah |
| 17:06 | <@froztbyte> | I have a couple of things to be replaced/purchased/etc in the near future |
| 17:06 | <@froztbyte> | soooooon. |
| 17:07 | | Syka [the@Nightstar-71fea65c.iinet.net.au] has joined #code |
| 17:07 | < Syka> | SIGH |
| 17:07 | < Syka> | apparently Telstra is having SEVERE INABILITIES TO ISP |
| 17:07 | < Syka> | and there's state-wide outages |
| 17:07 | <@froztbyte> | heeee |
| 17:07 | < Syka> | i called them up |
| 17:08 | < Syka> | and they were like yeah |
| 17:08 | < Syka> | telstra's done borked it |
| 17:08 | < Syka> | but now it's all good |
| 17:08 | < Syka> | 3G was working fine |
| 17:50 | < Xon> | Syka, it's been that way for most of the week |
| 17:51 | < Syka> | :( |
| 17:51 | < Xon> | Business connections with Telsta are not actually garrentied for quality btw |
| 17:52 | < Xon> | and they'll basicly laugh in your face if copper services are brought up and faults |
| 17:52 | < Syka> | eh, when I worked at my old work they were pretty good about it |
| 17:53 | < Syka> | but then again we were government |
| 17:53 | < Xon> | last 4-5 years it's gotten rather bad if you are a small ISP |
| 17:54 | | Kindamoody is now known as Kindamoody|out |
| 18:43 | | * ToxicFrog|W`rkn discovers LZ4 compression |
| 19:17 | <@celticminstrel> | ...why is instanceof Array returning false when this is clearly being printed out as an array. |
| 19:20 | <&ToxicFrog|W`rkn> | Because it's not an array but tostrings as one? |
| 19:21 | <@celticminstrel> | But it should be an array. It was assigned as an array. |
| 19:21 | <&ToxicFrog|W`rkn> | language? |
| 19:22 | <@celticminstrel> | JS |
| 19:22 | <@celticminstrel> | In the console, instanceof Array returns true. |
| 19:22 | <@Tamber> | inb4 "Well, there's your problem" |
| 19:22 | | * celticminstrel smacks Tamber. |
| 19:22 | | * Tamber ow |
| 19:22 | <@Tamber> | :D |
| 19:24 | < ErikMesoy> | Some people, when they have a problem, think "I know, I'll divide it into smaller and more manageable problems." Now they have several problems. |
| 19:24 | <@Tamber> | The Trouble with Problems. |
| 19:28 | <@celticminstrel> | Seems to be somehow related to the fact that it was assigned from the console. |
| 19:29 | < Syka> | dammit |
| 19:29 | < Syka> | I was too late to say what Tamber inb4'd |
| 19:29 | < Syka> | says syka, wrangling with her own terrible JS |
| 19:32 | | Kindamoody|out is now known as Kindamoody |
| 19:42 | | Kindamoody is now known as Kindamoody[zZz] |
| 20:15 | < [R]> | celticminstrel: node.js at all? |
| 20:15 | <@celticminstrel> | No. |
| 20:15 | <@celticminstrel> | Haven't tried that. |
| 20:15 | <@celticminstrel> | I'm doing client-side stuff. |
| 20:15 | < [R]> | Oh, because that's a known issue with it. |
| 20:16 | < [R]> | But, the correct way to detect arrays is Array.isArray() |
| 20:16 | <@celticminstrel> | I imagine this is an issue with Firefox's web console, then. |
| 20:16 | <@celticminstrel> | Oh, I see. |
| 20:16 | <@celticminstrel> | Is that defined on Object or something? |
| 20:17 | < [R]> | What is? |
| 20:17 | <@celticminstrel> | isArray() |
| 20:17 | < [R]> | No. |
| 20:17 | < [R]> | If it was, it'd be Object.isArray(variableToCheck) |
| 20:17 | <@celticminstrel> | Oh wait, it's a method in Array, not a method in Array.prototype? |
| 20:17 | < [R]> | Correct |
| 20:18 | <@celticminstrel> | So it'd be Array.isArray(variable). |
| 20:18 | < [R]> | Yes |
| 20:22 | <@Azash> | https://fbcdn-sphotos-e-a.akamaihd.net/hphotos-ak-ash4/283611_504220759632935_67 4989208_n.jpg |
| 20:25 | < AnnoDomini> | http://www.npr.org/blogs/alltechconsidered/2013/05/15/184223110/new-rifle-on-sal e?ft=1&f=1006&utm_source=dlvr.it&utm_medium=tw&utm_campaign=nprbusiness |
| 20:29 | | Turaiel is now known as Turaiel[Offline] |
| 21:24 | <&McMartin> | ErikMesoy: Floaters can teleport |
| 21:24 | <&McMartin> | Like, I've had my scouts see them do it. |
| 21:25 | <&McMartin> | It's very annoying, in an "oh shit half my guys are flanked" sense, but as a rule at least on Normal difficulty they fail to teleport into cover and aren't allowed to fire afterwards |
| 21:31 | < ErikMesoy> | McMartin: Oh, floater liftoff is legit. But on further inspection it may be more of a Schrodinger's Alien, whose position is unknown until you walk near one of the two locations it may be. |
| 21:31 | <&McMartin> | Ah |
| 21:31 | < ErikMesoy> | I think it may be a clever anti-savescumming measure. |
| 21:31 | <&McMartin> | Oh, yeah, there's some of that |
| 21:31 | <&McMartin> | I've been experimenting with those a little~ |
| 21:32 | | * McMartin had his first Shameful Savescum last night. |
| 21:32 | <&McMartin> | I accidentally murdered the shit out of an alien I wanted to capture with like five overwatch shots~ |
| 21:32 | < ErikMesoy> | OTOH, it does terrible things to my immersion when walking left will reveal aliens on the left, walking right from the previous savepoint will reveal aliens on the right, and afterwards there are no further aliens of that type to be found. |
| 21:33 | <&McMartin> | Given that you are modifying causality itself here your immersion must admit the presence of a hostile meddling God~ |
| 21:33 | <&McMartin> | Since there's at least one benevolent meddling God~ |
| 21:34 | < ErikMesoy> | McMartin: Oh, there's certainly a hostile God involved (meddling has yet to be proven) when I can walk a unit forwards, triggering a spawn of 3 chryssalids, who use their "reaction" to walk into melee with that unit. |
| 21:35 | <&McMartin> | Heh. This is why I scout with rookies >_> |
| 21:35 | <&McMartin> | Still, Chryssalids. |
| 21:35 | < ErikMesoy> | Move-on-spawning gameplay is a bitch that way |
| 21:35 | <&McMartin> | If they weren't stependously lethal cheaty bastards they wouldn't be Chryssalids >_> |
| 21:36 | < ErikMesoy> | Now with twice as much terror! |
| 21:37 | < ErikMesoy> | Since remake zombies burst on a timer rather than on death, which means zombified civilians in terror missions... |
| 21:41 | <&McMartin> | I fully approve of things that shock old-timers out of the complacency. |
| 21:42 | <&McMartin> | Like pairing them with the guys that used to be the weakest enemies in the game by a gigantic margin, and buffing the Hell out of those too~ |
| 21:42 | < ErikMesoy> | Pairing? My impression is that the remake has thrown that out of the window. |
| 21:42 | <&McMartin> | Hm. OK, it's May 1st for me |
| 21:42 | < ErikMesoy> | My most recent terror mission mixed together chrysalids, mutons, sectopods and heavy floaters. |
| 21:42 | <&McMartin> | Ah, OK |
| 21:42 | <&McMartin> | Maybe I just don't have enough variety in enemies~ |
| 21:45 | < ErikMesoy> | I snickered when I saw what they'd done with mutons (pinball!) and then what ELSE they'd done with mutons. |
| 21:45 | < ErikMesoy> | Also the Sectopod is just completely unfair now. |
| 21:46 | < ErikMesoy> | Yay for ghost armor and returning unfairness. :P |
| 21:47 | <&ToxicFrog|W`rkn> | I should play XCOM at some point. |
| 21:48 | < ErikMesoy> | You should play X-COM first, and then join the edition war. |
| 21:48 | <&ToxicFrog|W`rkn> | I have in fact played and beaten X-COM and TFTD. |
| 21:49 | < ErikMesoy> | Ah, carry on then. |
| 21:49 | <&McMartin> | I've only faced Mutons once so far |
| 21:49 | <&McMartin> | (I just got the "Off My Planet" achievement) |
| 21:51 | <&ToxicFrog|W`rkn> | (and played and quit in disgust Apocalypse~) |
| 21:52 | < ErikMesoy> | Funny, I liked Apocalypse but got disgusted by TFTD. |
| 21:54 | <&ToxicFrog|W`rkn> | The bits I hated about TFTD were the bits that were in every single mission in Apocalypse. |
| 21:54 | < ErikMesoy> | Such as... |
| 21:56 | | VirusJTG [VirusJTG@Nightstar-09c31e7a.sta.comporium.net] has joined #code |
| 21:56 | < ErikMesoy> | Apocalypse had best payback. Take laser bolters or some other endless-ammo weapon, attack Sirius base on realtime, sit in corner, select squad, hold fire button until base is razed. (Yes, yes, realtime sucks, but it lets you fire continuously.) |
| 21:56 | <&ToxicFrog|W`rkn> | huge levels containing hundreds of tiny rooms with aliens hiding in 1% of them, i.e. cruise ship missions in TFTD |
| 21:56 | <&ToxicFrog|W`rkn> | That is every single mission on Apocalypse, as far as I can tell, and you don't start out with enough firepower that leveling the entire building is a possible response. |
| 21:58 | < ErikMesoy> | Eh, you get infinite-ammo weapons fairly early |
| 21:58 | <&ToxicFrog|W`rkn> | Or I could just play the original again, which doesn't have missions like that :P |
| 21:59 | < ErikMesoy> | Well, not as many. :p |
| 21:59 | < ErikMesoy> | It did have terror missions where I ended up scouring the third floor of a house to find the last floater |
| 22:02 | <&ToxicFrog|W`rkn> | Yeah, but even house-heavy terror missions are way less annoying than cruise ships. |
| 22:02 | <&ToxicFrog|W`rkn> | And more amenable to rockets. |
| 22:07 | < ErikMesoy> | I think what alleviated Apocalypse there was squad movement. |
| 22:07 | < ErikMesoy> | Hunting down the last alien + reaction stops means there's much less risk to dragging your team around by 4s. |
| 22:40 | | Turaiel[Offline] is now known as Turaiel |
| 23:03 | | Turaiel is now known as Turaiel[Offline] |
| 23:32 | | ErikMesoy is now known as ErikMesoy|sleep |
| --- Log closed Sat May 18 00:00:06 2013 |