Tue Sep 8 21:08:19 2009
[imud_code]
Aidil@Way of the Force: so you can get extra guarantees where needed.
Tue Sep 8 21:09:11 2009
[imud_code]
Silenus@Dead Souls Dev: in my unoptimized bytecode generation for LPC (actually llvm calls it bitcode)
Tue Sep 8 21:09:25 2009
[imud_code]
Quixadhal@Bloodlines: SOrry, gold stars are licensed seperately from Sun. That'll be $699.
Tue Sep 8 21:10:09 2009
[imud_code]
Silenus@Dead Souls Dev: i get alot of stuff like load type tag, check tag(branch), load value for each load.
Tue Sep 8 21:10:17 2009
[imud_code]
Cratylus@Dead Souls Dev: theres supposed to be an option in there
Tue Sep 8 21:10:26 2009
[imud_code]
Quixadhal@Bloodlines: Oh, and I'm sorry to hear about you infecting your nice sparc hardware with Solardog. I'll say a prayer to the BSD gods to help you find the light.
Tue Sep 8 21:11:11 2009
[imud_code]
Silenus@Dead Souls Dev: i didnt see too many defines for options in the parser.c either when i looked at it recently.
Tue Sep 8 21:11:22 2009
[imud_code]
Aidil@Way of the Force: solaris cccccccccccccccccccooooooooooooooooonnnnnnnnnnnnnnnnteeeeeeeeeeeexxxxxxxxxxxxxxxxttttttttt switch :P
Tue Sep 8 21:11:42 2009
[imud_code]
Cratylus@Dead Souls Dev: it's either undocumented and tricky or it was never there
Tue Sep 8 21:12:30 2009
[imud_code]
Cratylus@Dead Souls Dev: "It's worth noting that many people find this unfriendly, and the parser does have an option to automatically use the first match instead of generating an error message."
Tue Sep 8 21:12:38 2009
[imud_code]
Cratylus@Dead Souls Dev: from http://discworld.atuin.net/lpc/about/articles/natural_command_handling.html
Tue Sep 8 21:14:44 2009
[imud_code]
Cratylus@Dead Souls Dev: they seem to be talking about the nlp package
Tue Sep 8 21:15:05 2009
[imud_code]
Ideysus@ShadowMUDii: Hmm. Its probably a sign boot takes too long when I start trying to compose haikus about how annoying sparcs are during reboots...
Tue Sep 8 21:20:34 2009
[imud_code]
Wodan@Discworld: oh, i wouldn't say that, i've SEEN our parser :)
Tue Sep 8 21:21:23 2009
[imud_code]
Silenus@Dead Souls Dev: dgd's parse_string probably makes things quite a bit easier.
Tue Sep 8 21:22:52 2009
[imud_code]
Wodan@Discworld: no idea if that's any use for user input parsing
Tue Sep 8 21:25:00 2009
[imud_code]
Ideysus@ShadowMUDii: Huh. parse_string is a proper parser. That actually looks quite useful
Tue Sep 8 21:25:43 2009
[imud_code]
Ideysus@ShadowMUDii: I usually seem to end up using reg_assoc to tokenise and then hack something together...
Tue Sep 8 21:26:37 2009
[imud_code]
Silenus@Dead Souls Dev: wonder if there is some open source tomita style LR out there which one could link into fluffos.
Tue Sep 8 21:26:41 2009
[imud_code]
Aidil@Way of the Force: parse_string is a very versitile tool :)
Tue Sep 8 21:27:34 2009
[imud_code]
Aidil@Way of the Force: hmm. you could try mailing Dworkin about that, but as is, nope, license won't allow that.
Tue Sep 8 21:28:19 2009
[imud_code]
Ideysus@ShadowMUDii: Okay. Is there a way to forcibly break out of the solaris installation? I installed once and it went wrong and dropped me at a shell prompt, from which I could fdisk. Now I -want- to be dropped back at the shell...
Tue Sep 8 21:28:27 2009
[imud_code]
Aidil@Way of the Force: about 8 lines of text, but restricts modifications and deriving from it.
Tue Sep 8 21:29:22 2009
[imud_code]
Quixadhal@Bloodlines: Unless the hardware is so old that you'd need to worry about the heads crashing into the platters....
Tue Sep 8 21:29:49 2009
[imud_code]
Ideysus@ShadowMUDii: But that won't drop me at the shell prompt in the installer...
Tue Sep 8 21:29:56 2009
[imud_code]
Quixadhal@Bloodlines: I forgot about L1-A... good old forth console!
Tue Sep 8 21:30:00 2009
[imud_code]
Aidil@Way of the Force: I distribute what amounts to a fork, which is possible, but requires that I also provide a way to regain the unmodified version.
Tue Sep 8 21:30:18 2009
[imud_code]
Cratylus@Dead Souls Dev: i can understand why you dont want to power cycle
Tue Sep 8 21:30:44 2009
[imud_code]
Cratylus@Dead Souls Dev: the installer isnt really meant to be interrupted in a way it doesnt understand
Tue Sep 8 21:30:54 2009
[imud_code]
Cratylus@Dead Souls Dev: if you cant F5 out, you'll need to kick it
Tue Sep 8 21:31:10 2009
[imud_code]
Cratylus@Dead Souls Dev: he good news is a full power cycle isnt necessary
Tue Sep 8 21:31:38 2009
[imud_code]
Cratylus@Dead Souls Dev: if yer at the serial console, send it a break signal
Tue Sep 8 21:32:05 2009
[imud_code]
Ideysus@ShadowMUDii: The shell I want is /bin/sh on the installer disk.
Tue Sep 8 21:32:09 2009
[imud_code]
Cratylus@Dead Souls Dev: but at least you dont need a full power cycle
Tue Sep 8 21:32:19 2009
[imud_code]
Ideysus@ShadowMUDii: But that'd just put me back in the installer...
Tue Sep 8 21:33:05 2009
[imud_code]
Cratylus@Dead Souls Dev: you can then mount the sys disk and hax your root passwd
Tue Sep 8 21:33:43 2009
[imud_code]
Ideysus@ShadowMUDii: Oh. I already got the root password... I hexedited the original disks. I am just trying to use solaris's fdisk on the disks.
Tue Sep 8 21:33:47 2009
[imud_code]
Quixadhal@Bloodlines: Oooo, fancy. I was expecting to /dev/ string instead of cdrom. User friendly consoles these days.
Tue Sep 8 21:34:07 2009
[imud_code]
Aidil@Way of the Force: re: Wodan.. hmm? yes.. but you can always get the matching unmodified version from the same distribution source :)
Tue Sep 8 21:34:11 2009
[imud_code]
Cratylus@Dead Souls Dev: you could try the physical paths but usually the devalias is set
Tue Sep 8 21:34:42 2009
[imud_code]
Cratylus@Dead Souls Dev: anyway, you dont need to use fdisk on sparc using scsi disks
Tue Sep 8 21:34:48 2009
[imud_code]
Wodan@Discworld: surely i can just change it to fit fluffos and point to the original? perhaps i should read the bloody license :)
Tue Sep 8 21:35:02 2009
[imud_code]
Ideysus@ShadowMUDii: The installer complained about the disks not being formatted...
Tue Sep 8 21:35:25 2009
[imud_code]
Cratylus@Dead Souls Dev: it'll prompt you to put a su label on there
Tue Sep 8 21:35:27 2009
[imud_code]
Aidil@Way of the Force: but in this case, you may want to contact Dworkin about it also.
Tue Sep 8 21:35:32 2009
[imud_code]
Cratylus@Dead Souls Dev: you'll need to know what yer doing for fdisk
Tue Sep 8 21:35:35 2009
[imud_code]
Raudhrskal@Dead Souls Dev: the prom console uses /dev paths at all?
Tue Sep 8 21:35:39 2009
[imud_code]
Raudhrskal@Dead Souls Dev: thought you use some kinda scsi bus addresses...
Tue Sep 8 21:36:00 2009
[imud_code]
Cratylus@Dead Souls Dev: you can actually cd down the device trees in the prom
Tue Sep 8 21:36:18 2009
[imud_code]
Cratylus@Dead Souls Dev: that's how you get info on mac addresses and whatnot
Tue Sep 8 21:36:38 2009
[imud_code]
Ideysus@ShadowMUDii: The weird openboot forth thing has the potential to be very cool if I could actually understand it.
Tue Sep 8 21:36:54 2009
[imud_code]
Ideysus@ShadowMUDii: And if it didn't keep hanging on things like probe-scsi :-(
Tue Sep 8 21:37:19 2009
[imud_code]
Cratylus@Dead Souls Dev: yer proly missing a terminator somewhere, or have a bad device
Tue Sep 8 21:37:43 2009
[imud_code]
Quixadhal@Bloodlines: Be thankful your battery isn't dead. Then you get to type in the hex code to reprogram the MAC address before you can boot. (Sparc ELC -- hasn't been on for a few years).
Tue Sep 8 21:39:29 2009
[imud_code]
Quixadhal@Bloodlines: http://www.squirrel.com/squirrel/sun-nvram-hostid.faq.html
Tue Sep 8 22:00:23 2009
[imud_code]
Cratylus@Dead Souls Dev: i didnt need you ugly bags of mostly water ANYWAY
Tue Sep 8 22:21:30 2009
[imud_code]
Ideysus@ShadowMUDii: Battery. (chips in batter, batteries in chips... I get it...)
Tue Sep 8 22:22:31 2009
[imud_code]
Quixadhal@Bloodlines: If you call now, we'll send you TWO Embedded waffle irons! Just pay shipping on each.
Tue Sep 8 22:25:15 2009
[imud_code]
Cratylus@Dead Souls Dev: i decided to just hit all the controller buttons at once instead of doing the combos like the booklet says
Tue Sep 8 22:25:41 2009
[imud_code]
Cratylus@Dead Souls Dev: whether it introduces subtle errors into the driver at runtime, that's for smart people to determine
Tue Sep 8 22:27:04 2009
[imud_code]
Aidil@Way of the Force: sorry, I'm already occupied.. and I believe Sil is off for sleep.
Tue Sep 8 22:27:19 2009
[imud_code]
Aidil@Way of the Force: which is not such a bad idea actually.. :)
Tue Sep 8 22:27:29 2009
[imud_code]
Aidil@Way of the Force: good whatever part of day it is for you.
Tue Sep 8 22:28:57 2009
[imud_code]
Wodan@Discworld: did anyone write fluffos compatible layers for dgd yet?
Tue Sep 8 22:29:44 2009
[imud_code]
Aidil@Way of the Force: there is an example happycode implementation.
Tue Sep 8 22:30:04 2009
[imud_code]
Ideysus@ShadowMUDii: I thought the (::) simulation thing didn't actually work as a closure?
Tue Sep 8 22:30:25 2009
[imud_code]
Aidil@Way of the Force: you can preprocess lpc code and recompile it into the dgd dialect.
Tue Sep 8 22:30:47 2009
[imud_code]
Wodan@Discworld: i'd love to compare dgd speed/memory usage for a fluffos lib :)
Tue Sep 8 22:31:06 2009
[imud_code]
Quixadhal@Bloodlines: I would think it would be... writing a DGD compatibility layer for FluffOS would be a bit harder.
Tue Sep 8 22:31:22 2009
[imud_code]
Aidil@Way of the Force: that is about syntax, what the underlying implementation really does is another question.
Tue Sep 8 22:31:41 2009
[imud_code]
Quixadhal@Bloodlines: Getting FluffOS to mimic the whole statedump thing might be a bit of a pain. :)
Tue Sep 8 22:32:07 2009
[imud_code]
Aidil@Way of the Force: but besides Wodan's comment, which is an interesting experiment.. whats the point :)
Tue Sep 8 22:32:53 2009
[imud_code]
Ideysus@ShadowMUDii: Assembly lets you implement a language which does closures... but I don't think it could do them natively?
Tue Sep 8 22:32:57 2009
[imud_code]
Wodan@Discworld: well, if it's faster and it works, i'd drop fluffos :) so you'd get more users
Tue Sep 8 22:33:20 2009
[imud_code]
Quixadhal@Bloodlines: Actually, a preprocessor to convert a FluffOS-based lib into a DGD-based lib might be interesting. It would have to deal with closures at the very least.
Tue Sep 8 22:34:28 2009
[imud_code]
Aidil@Way of the Force: heh. well, it could be made to work probably, but if that would really be faster.. I don't know.. :)
Tue Sep 8 22:34:30 2009
[imud_code]
Ideysus@ShadowMUDii: Actually, I guess mudos doesn't do closures properly...
Tue Sep 8 22:35:29 2009
[imud_code]
Aidil@Way of the Force: iirc, it is always possible, just not always practical, to rewrite a closure in a language that does not support closures natively.
Tue Sep 8 22:35:31 2009
[imud_code]
Ideysus@ShadowMUDii: I mean, I sometimes have to resort to (: ... whatever[0] ... :) using whatever[0], a single element array, since 'whatever' can't be modified directly.
Tue Sep 8 22:35:52 2009
[imud_code]
Aidil@Way of the Force: it may just be a lot of work, or just be impractical for humans, but that doesn't hinder a machine at all.
Tue Sep 8 22:36:52 2009
[imud_code]
Aidil@Way of the Force: you are looking at syntax, but that is something the recompile can deal with, after which all that matters is if it can be rewritten into a language not supporting closures natively.
Tue Sep 8 22:36:54 2009
[imud_code]
Ideysus@ShadowMUDii: Well, yeah closures don't let you create anything a turing machines doesn't...
Tue Sep 8 22:37:40 2009
[imud_code]
Aidil@Way of the Force: since that question can be answered with yes, you are in theory able to simulate that completely.
Tue Sep 8 22:38:03 2009
[imud_code]
Ideysus@ShadowMUDii: But since mudos doesn't actually let you bind to a variable on the stack ( $(...) doesn't count)...
Tue Sep 8 22:38:54 2009
[imud_code]
Raudhrskal@Dead Souls Dev hacks root at Aidil's workstation and installs a speech synth.
Tue Sep 8 22:41:26 2009
[imud_code]
Quixadhal@Bloodlines: So, who likes setting up RAID arrays in weird configurations? :)
Tue Sep 8 22:45:00 2009
[imud_code]
Quixadhal@Bloodlines: I have 4 drives. My raid bios (sucky onboard) will let me define 2 logical disks, a raid 10 set across all 4 disks, but only part of the space, and a raid 0 set across the remainder for data I don't care about. Is that a good idea? IE: if a disk dies, can I still rebuild the raid 10 part and just restore or lose the raid 0 part?
Tue Sep 8 22:45:02 2009
[imud_code]
Ideysus@ShadowMUDii: My laptop has a raid 1 on two asymmetrically sized small SSDs if that counts?
Tue Sep 8 22:45:06 2009
[imud_code]
Sys@BlackHole: I've alawys thought you need 8 disks in a box, so you have a 5+1 as the main array and a mirror pair for /boot but that's not wierd
Tue Sep 8 22:46:03 2009
[imud_code]
Ideysus@ShadowMUDii: Why can't you put /boot on the 5+1? its generaly not large and generally not accessed much after boot...
Tue Sep 8 22:46:24 2009
[imud_code]
Ideysus@ShadowMUDii: I mean, on the same physical disks, but before it...
Tue Sep 8 22:47:10 2009
[imud_code]
Sys@BlackHole: I'd actually have more than just /boot I'd have the o/s on there, and I prefer mirrored o/s disks for reasons I've long forgotten
Tue Sep 8 22:47:50 2009
[imud_code]
Sys@BlackHole: it also meanns you can have a pair of smaller/faster or more reliable disks as the mirror pair
Tue Sep 8 22:48:14 2009
[imud_code]
Ideysus@ShadowMUDii: Well, 5 is slow, so depending on the application a mirrored OS may make more sense...
Tue Sep 8 22:49:19 2009
[imud_code]
Quixadhal@Bloodlines: Hmmmm, I'm thinking striping half the disks as raid 0 would be nice for games, since 99% of the data is easily restored anyways. Just seems a bit odd to mix types on the same disks. :)
Tue Sep 8 22:49:21 2009
[imud_code]
Sys@BlackHole: with flash memory being so small/cheap you could have 1/4 TB of buffer
Tue Sep 8 22:49:52 2009
[imud_code]
Raudhrskal@Dead Souls Dev: In 5, even if you change a single byte in a block, you need to recompute that bloack AND the parity.
Tue Sep 8 22:49:53 2009
[imud_code]
Sys@BlackHole: I know what you mean about mixing types, I felt wierd the first time I did it, but it seems to work ok
Tue Sep 8 22:50:01 2009
[imud_code]
Ideysus@ShadowMUDii: I thought raid 5 write performance was bad... Hmm. My memory is fuzzy... I guess it has to read every time it writes...
Tue Sep 8 22:50:06 2009
[imud_code]
Raudhrskal@Dead Souls Dev: That will require reading the connected blocks from all other drives too.
Tue Sep 8 22:50:44 2009
[imud_code]
Sys@BlackHole: good point, but surely fast processors and big buffers on yoru RAID card can mask that for any given data volume
Tue Sep 8 22:51:01 2009
[imud_code]
Raudhrskal@Dead Souls Dev: you need to read the stripe from all the drives, change the data, recompute parity, store the modified parts of the stripe and the new parity.
Tue Sep 8 22:51:13 2009
[imud_code]
Quixadhal@Bloodlines laughs at the idea of buffers on a raid card (as he looks at his onboard raid controller ans sniffs).
Tue Sep 8 22:52:06 2009
[imud_code]
Ideysus@ShadowMUDii: Of course, I only use software raid, which is probably a performance hit in itself :-)
Tue Sep 8 22:52:09 2009
[imud_code]
Sys@BlackHole: no I suppose not, so for Fast performance mirroring is the way forward
Tue Sep 8 22:53:25 2009
[imud_code]
Ideysus@ShadowMUDii: I like being able to actually pull my disks out and move them to another machine :-P
Tue Sep 8 22:53:44 2009
[imud_code]
Ideysus@ShadowMUDii: If I had a bunch of identical machines I guess I'd reconsider...
Tue Sep 8 22:54:14 2009
[imud_code]
Sys@BlackHole: I've not played with it yet, but will as soon as the new server is ordered
Tue Sep 8 22:54:51 2009
[imud_code]
Quixadhal@Bloodlines: Ok, interesting question... does the raid controller store the configuration data in nvram, or on each disk involved in the set? Just curious...
Tue Sep 8 22:56:07 2009
[imud_code]
Quixadhal@Bloodlines: Yeah, I suppose it doesn't matter much, unless you lose all disks involved, but thinking of mixing types made me wonder.
Tue Sep 8 22:56:45 2009
[imud_code]
Cratylus@Dead Souls Dev: if the controllers have cache, it's worth using the cache sometimes
Tue Sep 8 22:57:22 2009
[imud_code]
Quixadhal@Bloodlines: ZFS.... isn't that still buggy enough that even Microsoft backed away from using it?
Tue Sep 8 22:57:59 2009
[imud_code]
Quixadhal@Bloodlines: They were originally going to stick it in Vista, but ran off scampering and I see no signs of it in 7. Too bad, it sounded really nice.
Tue Sep 8 22:58:25 2009
[imud_code]
Cratylus@Dead Souls Dev: ms decisions on what to implement do not always relate to the brokenness of the product
Tue Sep 8 22:58:43 2009
[imud_code]
Cratylus@Dead Souls Dev: but rather their advantage in the market
Tue Sep 8 22:58:51 2009
[imud_code]
Quixadhal@Bloodlines: No, I'm talking about "news" I heard about 4 years ago. It may have been out of someone else's ass though.
Tue Sep 8 22:59:03 2009
[imud_code]
Sys@BlackHole: never a truer word said Crat, and broken software is better in the market
Tue Sep 8 22:59:37 2009
[imud_code]
Sys@BlackHole: certainly evry techie I spoke to who used zfs said it was loverly
Tue Sep 8 23:00:14 2009
[imud_code]
Quixadhal@Bloodlines: Cool. Maybe I'll use it next time I have to move my linux server to new hardware. I expect the next hardware failure to be in about 2 years.
Tue Sep 8 23:00:52 2009
[imud_code]
Sys@BlackHole: but on modern hardware you can install opensolaris on zfs and then run linux in a vm
Tue Sep 8 23:01:00 2009
[imud_code]
Quixadhal@Bloodlines: Probably a CPU fan this time... power supply and drives already failed in the last 5 years so the CPU fan is due.
Tue Sep 8 23:01:02 2009
[imud_code]
Cratylus@Dead Souls Dev: but people put their production servers on it
Tue Sep 8 23:01:38 2009
[imud_code]
Cratylus@Dead Souls Dev: production as in, their livelihood depends on it
Tue Sep 8 23:02:08 2009
[imud_code]
Quixadhal@Bloodlines: I might need a bigger hard drive though... will Solardog run nicely in a software raid 1 mirror of two 20g drives?
Tue Sep 8 23:02:19 2009
[imud_code]
Sys@BlackHole: tho your proof is weak, many people's livleyhoods depend on software that is broken
Tue Sep 8 23:02:41 2009
[imud_code]
Cratylus@Dead Souls Dev: i agree my argument seems weak on that point
Tue Sep 8 23:02:55 2009
[imud_code]
Zaphod@Dead Souls Dev: joyent is a pretty good example of a joint doing fancy production things on zfs
Tue Sep 8 23:08:07 2009
[imud_code]
Ideysus@ShadowMUDii: Hmm. So can you tell zfs 'mirror this data over all disks' or 'don't mirror this at all, stripe it over all disks'?
Tue Sep 8 23:08:50 2009
[imud_code]
Cratylus@Dead Souls Dev: bleeb so, except maybe boot disks which default to mirroring
Tue Sep 8 23:09:57 2009
[imud_code]
Zaphod@Dead Souls Dev: Only mirrored pools and pools with one disk are supported. No RAID-Z or unreplicated pools with more than one disk are supported.
Tue Sep 8 23:10:13 2009
[imud_code]
Ideysus@ShadowMUDii: Oh. Cool. Regarding that earlier raid5 thing, RAID-Z works around it by only ever writing a full stripe
Tue Sep 8 23:10:56 2009
[imud_code]
Ideysus@ShadowMUDii: ZFS does too much for me to comprehend fully. :-(
Tue Sep 8 23:11:36 2009
[imud_code]
Quixadhal@Bloodlines: Will it cook me eggs and bacon with rye toast?
Tue Sep 8 23:11:52 2009
[imud_code]
Zaphod@Dead Souls Dev: altho, it works rilly nicely in an xvm (Xen) domain
Tue Sep 8 23:13:01 2009
[imud_code]
Ideysus@ShadowMUDii: Huh. Its not clustered? Multiple hosts can't access it at once?
Tue Sep 8 23:13:28 2009
[imud_code]
Ideysus@ShadowMUDii: Doesn't that take a lot of the advantage out of having zfs over multiple machines?
Tue Sep 8 23:39:52 2009
[imud_code]
Ideysus@ShadowMUDii: Is there seriously no way to set raid up within the solaris installer?
Tue Sep 8 23:41:53 2009
[imud_code]
Cratylus@Dead Souls Dev: you add more than one disk, it automirrors
Tue Sep 8 23:42:49 2009
[imud_code]
Cratylus@Dead Souls Dev: if you wanna do software volumes with ufs you'll need to mess with lvm after install
Tue Sep 8 23:43:24 2009
[imud_code]
Cratylus@Dead Souls Dev: but that will be more work than you expect if you dont reserve a slice now for metastate databases
Tue Sep 8 23:43:39 2009
[imud_code]
Cratylus@Dead Souls Dev: wtf dude, like it's my fault yer a noob
Tue Sep 8 23:44:53 2009
[imud_code]
Cratylus@Dead Souls Dev: but with hans reiser being a murderer now, suddenly reiserfs isnt default much anymore
Tue Sep 8 23:45:25 2009
[imud_code]
Raudhrskal@Dead Souls Dev: and still, if you dd a reiser parition to an image, store it on another reiser partition, and finally fsck it it'll try to merge the trees and kill everything.
Tue Sep 8 23:46:43 2009
[imud_code]
Ideysus@ShadowMUDii: Okay, how does this make sense: 'choose zfs or ufs. choose software to install. choose disks. choose layout '
Tue Sep 8 23:47:27 2009
[imud_code]
Ideysus@ShadowMUDii: Surely choosing filesystem should come after choosing disks... can you even have a mixed system? They didn't even show F3 as a back button...
Tue Sep 8 23:47:50 2009
[imud_code]
Ideysus@ShadowMUDii: Umm... I like the flashy lights everywhere?
Tue Sep 8 23:48:37 2009
[imud_code]
Cratylus@Dead Souls Dev: the disk customization process for zfs and ufs is very different
Tue Sep 8 23:49:01 2009
[imud_code]
Corrik@Lost Realms: When add_action() is enabled on FluffOS, is there an efun that returns all of the actions added to an object?
Tue Sep 8 23:49:03 2009
[imud_code]
Cratylus@Dead Souls Dev: and it's not unreasonable to consider disk selection to be part of disk customization
Tue Sep 8 23:50:22 2009
[imud_code]
Ideysus@ShadowMUDii: Well, software choice probably should separate the tasks... nevermind. ignore me, I've just been spoilt by the debian installer
Tue Sep 8 23:51:05 2009
[imud_code]
Raudhrskal@Dead Souls Dev: which's quite crazy in its own way, where you can back up at any moment but it silently resets the other parameters if you do it.
Tue Sep 8 23:51:55 2009
[imud_code]
Ideysus@ShadowMUDii: Isn't that intuitive. Huh, I guess its what you are used to.
Wed Sep 9 20:33:58 2009
[imud_code]
Benedick@WWC: -I- heard he was funding the Irish Republican Army.
Wed Sep 9 23:23:42 2009
[imud_code]
Silenus@Dead Souls Dev: do all the allocations in fluffos go through the custom allocators or are there exceptions?
Fri Sep 11 02:12:10 2009
[imud_code]
Ideysus@ShadowMUDii: Is 'foreach(x in y)' synonymous with 'foreacch(x in y+({}))'? That is, does foreach operate on a copy of the array, thus safely allowing modification of the array?
Fri Sep 11 02:13:33 2009
[imud_code]
Cratylus@Dead Souls Dev: ive always assumed that modifying x does not modify y
Fri Sep 11 02:14:02 2009
[imud_code]
Ideysus@ShadowMUDii: I'm quite sure modifying x is safe, I'm modifying y while in the array
Fri Sep 11 02:16:47 2009
[imud_code]
Ideysus@ShadowMUDii: Hmm. Oh wait... it doesn't work... Appending to the array doesn't affect the foreach, but modifying elements does
Fri Sep 11 03:43:08 2009
[imud_code]
Cratylus@Dead Souls Dev: copy() is an efun in fluff, negligible perf hit, and with belt and suspenders yer trousers wont go nowhere
Fri Sep 11 03:44:38 2009
[imud_code]
Tricky@andSapid: I think it is safe to transfer RtH over to v2.17.3-beta-rth now.
Fri Sep 11 03:45:29 2009
[imud_code]
Tricky@andSapid: Stopped the lockups by getting rid of the buggy LPC httpd.
Fri Sep 11 03:45:44 2009
[imud_code]
Ideysus@ShadowMUDii: I wish I could make references read-only...
Fri Sep 11 03:46:50 2009
[imud_code]
Ideysus@ShadowMUDii: Like, if m=ro(m) meant m[1]="x; errored...
Fri Sep 11 03:47:05 2009
[imud_code]
Kalinash@Fire and Ice: structures are a language concept, not a computer concept
Fri Sep 11 03:47:32 2009
[imud_code]
Kalinash@Fire and Ice: you can cast any memory location to a structure
Fri Sep 11 03:48:02 2009
[imud_code]
Kalinash@Fire and Ice: a structure is just offsets from a base pointer
Fri Sep 11 03:48:09 2009
[imud_code]
Cratylus@Dead Souls Dev: so if mp->ordinal is segfaulting, i can (void *)mp->ordinal ?
Fri Sep 11 03:48:24 2009
[imud_code]
Cratylus@Dead Souls Dev: without risking the actual contents of mp if it IS a struct?
Fri Sep 11 03:49:39 2009
[imud_code]
Tricky@andSapid: err... it isn't ever a struct. It's just a bit of memory that you reference with the struct.
Fri Sep 11 03:49:55 2009
[imud_code]
Kalinash@Fire and Ice: if mp->foo is crashing, then most likely the address of mp is invalid
Fri Sep 11 03:50:18 2009
[imud_code]
Kalinash@Fire and Ice: in C pointers (anything defined with a *) is just a memory location
Fri Sep 11 03:50:48 2009
[imud_code]
Kalinash@Fire and Ice: so at times your pointer is invalid... what is it's value when it crashes?
Fri Sep 11 03:51:04 2009
[imud_code]
Kalinash@Fire and Ice: if it's null, or 0, then tricky is right
Fri Sep 11 03:51:23 2009
[imud_code]
Kalinash@Fire and Ice: if the pointer value is 1 then it will also crash
Fri Sep 11 03:51:36 2009
[imud_code]
Kalinash@Fire and Ice: since you can't reference 0 page memory on modern OSes
Fri Sep 11 03:52:51 2009
[imud_code]
Kalinash@Fire and Ice: I didn't really like the 6502 that much, but respected it's proliferation
Fri Sep 11 03:53:58 2009
[imud_code]
Tricky@andSapid: Doub't you would have had the consoles of today without it.
Fri Sep 11 03:56:32 2009
[imud_code]
Cratylus@Dead Souls Dev: i'll make it if mp != NULL && mp->foo != NULL
Fri Sep 11 03:56:36 2009
[imud_code]
Tricky@andSapid: If with a malloc are you checking the return value?
Fri Sep 11 03:56:53 2009
[imud_code]
Kalinash@Fire and Ice: you have to make sure that mp is initialized to NULL and set back to NULL after you free it
Fri Sep 11 03:57:14 2009
[imud_code]
Kalinash@Fire and Ice: otherwise the NULL check won't matter since it will be an undefined value
Fri Sep 11 03:57:28 2009
[imud_code]
Cratylus@Dead Souls Dev: did i ever tell you about my cat's breath?
Fri Sep 11 03:57:37 2009
[imud_code]
Cratylus@Dead Souls Dev: the cool thing is the crash is pretty reproduceable
Fri Sep 11 03:58:08 2009
[imud_code]
Kalinash@Fire and Ice: tricky: probably, but just because the VCS, NES and TG-16 used it doesn't mean I had to like it :P
Fri Sep 11 03:58:19 2009
[imud_code]
Tricky@andSapid: also with a malloc are you doing sommat like mp = (structname_t *)malloc(sizeof(structname_s));
Fri Sep 11 03:58:23 2009
[imud_code]
Kalinash@Fire and Ice: the 6809 was the best 8-bit cpu produced
Fri Sep 11 03:59:10 2009
[imud_code]
Kalinash@Fire and Ice: i also liked the Z80 since it had the IN/OUT opcodes that used a seperate address space for interacing instead of having to use address space mapping
Fri Sep 11 03:59:19 2009
[imud_code]
Tricky@andSapid: Hmm... probably will have to check the specs on the 6809
Fri Sep 11 04:02:53 2009
[imud_code]
Cratylus@Dead Souls Dev: ok srsly, how do i check mp is a valid struct?
Fri Sep 11 04:03:09 2009
[imud_code]
Cratylus@Dead Souls Dev: there has to be some simple way to check
Fri Sep 11 04:03:24 2009
[imud_code]
Tricky@andSapid: Without knowing how you create it there is know way of knowing.
Fri Sep 11 04:04:16 2009
[imud_code]
Tricky@andSapid: As part of a global/local variable or a malloc?
Fri Sep 11 04:06:45 2009
[imud_code]
Tricky@andSapid: I'm guessing ... mp = malloc(sizeof(match_s));
Fri Sep 11 04:08:24 2009
[imud_code]
Tricky@andSapid: However, once malloced you check the return value for NULLness
Fri Sep 11 04:09:03 2009
[imud_code]
Tricky@andSapid: Before doing anything with it. Then free(mp) at the end.
Fri Sep 11 04:13:18 2009
[imud_code]
Echlori@The New Horizon: I haven't played around with ds...but are the arch commands secure in ds?
Fri Sep 11 04:21:01 2009
[imud_code]
Kalinash@Fire and Ice: does mp get assigned to an existing struct or is it supposed to be a new one?
Fri Sep 11 04:21:59 2009
[imud_code]
Cratylus@Dead Souls Dev: left undefined in the error cases, it seems
Fri Sep 11 04:22:31 2009
[imud_code]
Kalinash@Fire and Ice: so try this... change the line to: match_t *mp = NULL;
Fri Sep 11 04:22:50 2009
[imud_code]
Kalinash@Fire and Ice: then check for NULL before you use it... you should also assert it
Fri Sep 11 04:23:08 2009
[imud_code]
Kalinash@Fire and Ice: since it seems your change is causing a code path to execute that didn't previously
Fri Sep 11 04:23:42 2009
[imud_code]
Kalinash@Fire and Ice: it would be good to know if that's an intended or unintended consequence
Fri Sep 11 04:24:53 2009
[imud_code]
Kalinash@Fire and Ice: next step would be to send me the code and the repro steps and let me take a look locally ;)
Fri Sep 11 04:25:23 2009
[imud_code]
Kalinash@Fire and Ice: you're not running this on Solaris are you?
Fri Sep 11 04:29:56 2009
[imud_code]
Cratylus@Dead Souls Dev: interestingly, tricky's weird malloc solution seemed to be working
Fri Sep 11 04:30:05 2009
[imud_code]
Kalinash@Fire and Ice: well sure, you're assinging to valid memory
Fri Sep 11 04:30:07 2009
[imud_code]
Cratylus@Dead Souls Dev: but i'm pretty sure i screwed it up in some way that introduced errors
Fri Sep 11 04:30:11 2009
[imud_code]
Kalinash@Fire and Ice: but it doesn't sound like that's the intended behavior
Fri Sep 11 04:30:38 2009
[imud_code]
Cratylus@Dead Souls Dev: i added code, it disrupted the beek flow
Fri Sep 11 04:31:43 2009
[imud_code]
Tricky@andSapid: weird ... malloc ... solution. You should visit alt.c.faq for a real flame war with comments like that.
Fri Sep 11 04:31:58 2009
[imud_code]
Cratylus@Dead Souls Dev: nobody was going to do this thing, so it fell to me, the C newbie
Fri Sep 11 04:35:02 2009
[imud_code]
Tricky@andSapid: Yeah! That's the image I get of you typing with a baby.
Fri Sep 11 04:36:05 2009
[imud_code]
Cratylus@Dead Souls Dev: well hot diggity, i think that's got it
Fri Sep 11 04:36:29 2009
[imud_code]
Cratylus@Dead Souls Dev: ima go with the one that doesnt malloc tho
Fri Sep 11 04:36:36 2009
[imud_code]
Cratylus@Dead Souls Dev: since im pretty sure i'd screw that up
Fri Sep 11 04:36:45 2009
[imud_code]
Tricky@andSapid: Just to be sure, you AREN'T crediting us with you buggy code, right?
Fri Sep 11 05:55:11 2009
[imud_code]
Tricky@andSapid: I see a potential overflow problem in add_match(...) in parser.c
Fri Sep 11 05:56:18 2009
[imud_code]
Tricky@andSapid: There is no check for how big state->num_matches can get, however the global variable matches[] is hardcoded to 10.
Fri Sep 11 13:30:53 2009
[imud_code]
Cratylus@Dead Souls: so i left this mud running overnight witht he new parser and tons of bots and just as icing on the cake, tons of aggr npcs
Fri Sep 11 17:14:35 2009
[imud_code]
Kalinash@Fire and Ice: did you fix your parser crash last night?
Fri Sep 11 17:14:41 2009
[imud_code]
Zaphod@Dead Souls Dev basks in the sense of loving community around here
Fri Sep 11 17:14:44 2009
[imud_code]
Cratylus@Dead Souls Dev: 11 years i suffered under beek's parser oppression
Fri Sep 11 17:15:20 2009
[imud_code]
Kalinash@Fire and Ice: http://en.wikipedia.org/wiki/Dixie%27s_BBQ
Fri Sep 11 18:23:35 2009
[imud_code]
Cratylus@Dead Souls Dev: so i mangled the parser code in the dirver
Fri Sep 11 18:23:49 2009
[imud_code]
Cratylus@Dead Souls Dev: it did whati wanted, but it crashed about 10% of the time
Fri Sep 11 18:24:03 2009
[imud_code]
Cratylus@Dead Souls Dev: kali told me how to get rid of the crash
Fri Sep 11 18:24:31 2009
[imud_code]
Cratylus@Dead Souls Dev: but tricky's fix was easier to screw up
Wed Sep 16 02:05:00 2009
[imud_code]
Taffyd@Discworld: We are crashing a lot on FluffOS 2.17, FYI. If any of you guys are early adopters.
Wed Sep 16 02:05:15 2009
[imud_code]
Taffyd@Discworld: DW is going back to 2.16 seeing as Wodan is asleep:p
Wed Sep 16 02:06:01 2009
[imud_code]
Tricky@Rock the Halo: I ought to try my variant of 2.18-alpha on DS with lots of bots to test it.
Thu Sep 17 21:33:56 2009
[imud_code]
Ideysus@ShadowMUDii: Anybody got any good ideas for 'catch_unlogged' in mudos?
Thu Sep 17 21:34:48 2009
[imud_code]
Ideysus@ShadowMUDii: I am thinking something like make it a sefun catch_unlogged(mixed m){return catch(m);}
Thu Sep 17 21:35:16 2009
[imud_code]
Ideysus@ShadowMUDii: Then perhaps error_handler can identify catch_unlogged with call_stack?
Go to the top | Channel Index
