Conversation with #inferno at Mon Jan 17 02:53:10 2011 on powerman-asdf@irc.freenode.net (irc) (04:17:26) caerwyn left the room (quit: Ping timeout: 240 seconds). (04:39:19) caerwyn [~caerwyn@c-98-217-102-21.hsd1.ma.comcast.net] entered the room. (05:27:28) vsriniva1 is now known as vsrinivas (05:47:50) caerwyn left the room (quit: Quit: Colloquy for iPad - http://colloquy.mobi). (05:55:26) cbj1 left the room (quit: Ping timeout: 240 seconds). (06:54:47) base2design: still need an arm/linux emu? (06:54:57) base2design: (vsriniva1) (06:57:47) base2design: okay, maybe i've been up too long, but i can't make an SSL connection to a website (trying to write a little wget in limbo for https retrieval). any pointers beyond the man pages (i'm missing something and keep getting "hung i/o") when using security.m. (07:07:07) base2design: (no luck at all understanding the ssl3/tls stuff either (07:45:16) robot12 [~KAZZhilki@proxy10.ts.fujitsu.com] entered the room. (07:50:39) bvalek2 [c3e41404@gateway/web/freenode/ip.195.228.20.4] entered the room. (08:31:07) uriel: base2design: you might want to try Go, it has excellent crypto support (08:34:34) Gegemon [~ynv@mx1.airis.ru] entered the room. (09:41:24) mjl-: morning (09:43:12) mjl-: base2design: you could take a look at function "pushssl" in http://www.ueber.net/hg/http/files/tip/appl/lib/mhttp.b (09:43:35) mjl-: that's a http library, that package also has "hget.b", which is also a bit like wget (09:45:07) mjl-: the pushssl function takes an fd and an address (with ssl being spoken in the fd). and returns an fd from which you can read the raw data, i.e. with ssl "unpeeled". (09:53:40) robot121 [~KAZZhilki@proxy10.ts.fujitsu.com] entered the room. (09:55:00) robot12 left the room (quit: Ping timeout: 246 seconds). (09:56:10) robot12 [~KAZZhilki@proxy10.ts.fujitsu.com] entered the room. (09:58:17) robot121 left the room (quit: Ping timeout: 264 seconds). (09:59:00) robot121 [~KAZZhilki@proxy10.ts.fujitsu.com] entered the room. (09:59:00) mjl-: base2design: oh, another thing about pushssl (actually, the module that is used by it, so pushssl cannot solve it): the .read function of the ssl context acts as readn, so it blocks until all requested bytes have been read (or until eof). (10:00:33) mjl-: for hget, that forced me to put a "connection: close" header in. without it, the eof never comes and your reads won't return (unless you read the exact number of bytes you want, which you may not know if the content-length isn't set). (10:00:41) robot12 left the room (quit: Ping timeout: 264 seconds). (10:01:11) mjl-: also, it's ssl3. not tls. so also no protection against the mitm attack that was uncovered last year or the year before that. (10:14:31) mjl-: today on the menu: build emu on windows xp with the latest ms c compiler (10:40:49) mjl-: 2.4gb disk space required, cannot be on a network share... (11:17:53) robot12 [~KAZZhilki@proxy10.ts.fujitsu.com] entered the room. (11:18:41) robot121 left the room (quit: Ping timeout: 264 seconds). (11:20:35) robot121 [~KAZZhilki@proxy10.ts.fujitsu.com] entered the room. (11:21:56) robot12 left the room (quit: Ping timeout: 240 seconds). (11:24:44) robot121 left the room (quit: Ping timeout: 240 seconds). (11:30:18) bvalek2 left the room (quit: Ping timeout: 265 seconds). (11:32:05) sl left the room (quit: Ping timeout: 272 seconds). (11:33:50) robot12 [~KAZZhilki@proxy10.ts.fujitsu.com] entered the room. (12:31:47) robot121 [~KAZZhilki@proxy10.ts.fujitsu.com] entered the room. (12:32:55) robot12 left the room (quit: Ping timeout: 276 seconds). (12:38:55) sl [~sl@68-179-130-17.bsr-c9-d1.evv.dhcp.sigecom.net] entered the room. (12:53:43) GriffenJBS: mjl, thanks for trying those builds, the rxxx was the one you sent me, the r165 and r175 were from sourceforge, and the idds, isds were by local builds (12:55:11) mjl-: GriffenJBS: ok, so your builds didn't work for me either (13:05:42) mjl-: GriffenJBS: in the kernel configs, perhaps you still have "dldi"? and the "io*" in the link section? (13:29:57) mjl-: the Nt network fix is already in inferno-os (13:30:32) mjl-: i think i messed that up. i am sure i created that patch before some time, probably didn't attach to the patch in the bug tracker though :( (13:32:39) robot121 is now known as robot12 (13:38:56) vsrinivas: caerwyn's "scope" works fairly well :D (14:28:46) base2design left the room (quit: Ping timeout: 250 seconds). (14:51:17) base2design [~base2desi@97-80-161-40.dhcp.gwnt.ga.charter.com] entered the room. (14:51:45) base2design: mjl: thanks! once again, your example code "saves the day" (14:55:55) base2design: uriel: go looks very nice, but i need to stick with inferno for the moment... indeed, i would use it if the arm compiler didn't generate binaries that were quite so large (memory usage is about 2x that of inferno). (14:59:13) uriel: really? that sounds strange (14:59:37) uriel: if you take into account the Go runtime vs. the Inferno runtime, I doubt they are bigger (14:59:40) uriel: but I could be wrong (15:00:03) uriel: i'd talk with ken, he is the one hacking on the Go arm compilers lately (15:24:23) mjl-: hah, my accomplishment of the day so far: (15:24:34) mjl-: a tool that rewrites a .zip file so it is more streamable (15:24:58) mjl-: i.e. with the table of contents at the start of the file instead of at the end. too bad info-zip and 7z don't understand it (they suck). (15:27:28) base2design: the memory hit for inferno is justifiable because i can have several "apps" running under the same runtime. since each app carries the go runtime, this gets duped (unless there's some linux behind the scenes magic that i don't know about) (15:28:12) base2design: oh, so here' s stupid newbie question... when i run "hget", i get the message "dial net!10.0.1.1!443: '/net/net' file does not exist (/net/net/clone)" (15:28:38) base2design: i have a very minimal inferno installation so i've probably mis-config'ed something (15:29:08) base2design: my "/net" only has arp,ndp,tcp,udp (15:32:39) mjl-: run ndb/cs first (15:42:54) base2design: oops (16:23:37) btdn left the room (quit: Quit: leaving). (16:34:39) Fish-- left the room (quit: Quit: So Long, and Thanks for All the Fish). (16:34:46) Fish [~Fish@exo3753.pck.nerim.net] entered the room. (17:03:33) Gegemon left the room (quit: Quit: Leaving.). (17:41:22) robot12 left the room (quit: Quit: Leaving.). (17:52:31) GriffenJBS: mjl- what did you mean "still have", did I miss a setting or patch? (18:01:44) mjl-: well, you had to remove the dldi code from the kernel (18:01:51) mjl-: by modifying the kernel config file (18:01:57) mjl-: i was wondering if you had done that (18:02:15) GriffenJBS: no, I have not (18:02:29) GriffenJBS: it's strange the dldi code couldn't be used (18:03:58) mjl-: well, there is at least the difference in calling convention that makes it impossible to call dldi machine code without additional work. it can be fixed, should be, but it does not work now (18:04:32) mjl-: the kernel config files are ds,sds,dds (18:04:42) mjl-: the line with "dldi" should get a # in front of it (18:05:45) mjl-: and so should the lines in the "link" section (which starts with an unindented sole "link" on a line) that start with "io" (eg "iottds") (18:05:48) GriffenJBS: dldi was included in ds, dds, and sds (18:06:06) mjl-: yes, and it makes kernels not work (18:06:36) GriffenJBS: I haven't seen it written anywhere, what is the difference between ds, dds and sds? (18:07:35) GriffenJBS: sds says "all files in root" dds says "all files in sd/tf card", and ds has no comment (18:07:42) GriffenJBS: I (18:07:59) mjl-: you can diff them... (18:08:14) GriffenJBS: 'm assuming ds uses nitro, or the like, and the other two are differnt ... seems the same to me (18:08:16) mjl-: it's the kernel config, so it's a difference in drivers and files in the built-in root file system (18:09:39) GriffenJBS: I understand they're different profiles, I don't understand to source tree well enough to know the difference I'm looking at... guess time will fix that (18:09:49) mjl-: so, dds has few built-in files in the root file system (do a "diff ds dds"), because it's supposed to find the files on the sd card (18:10:38) mjl-: sds seems to be the opposite: as many files in the built-in root file system as wanted/possible (18:11:28) mjl-: this is the manual page describing the kernel config files: http://www.vitanuova.com/inferno/man/10/conf.html (18:13:41) GriffenJBS: thank you (18:45:19) GriffenJBS: ds had a link section, but nothing starting with io, commented out dldi under the dev section, rebuild, still just blank screens (18:57:06) base2design_ [~basedesig@166.137.15.15] entered the room. (18:58:49) bvalek2 [598516ea@gateway/web/freenode/ip.89.133.22.234] entered the room. (19:48:00) eno left the room (quit: Read error: Connection reset by peer). (19:52:40) eno [~eno@nslu2-linux/eno] entered the room. (20:44:55) base2design__ [~basedesig@97-80-161-40.dhcp.gwnt.ga.charter.com] entered the room. (20:48:11) base2design_ left the room (quit: Ping timeout: 255 seconds). (20:50:35) base2design__ left the room (quit: Quit: Colloquy for iPhone - http://colloquy.mobi). (20:56:54) bvalek2 left the room (quit: Quit: Page closed). (22:35:22) Fish- [~Fish@9fans.fr] entered the room. (23:25:20) base2design_ [~basedesig@166.137.14.32] entered the room. (23:25:51) base2design_ left the room (quit: Client Quit). (00:01:45) base2design: h (00:57:07) Fish- left the room (quit: Quit: So Long, and Thanks for All the Fish). (01:14:38) base2design left the room (quit: Ping timeout: 240 seconds). (01:16:43) caerwyn [~caerwyn@c-98-217-102-21.hsd1.ma.comcast.net] entered the room. (01:17:49) base2design [~base2desi@97-80-161-40.dhcp.gwnt.ga.charter.com] entered the room. (01:27:51) cbj1 [~none@c-98-217-102-21.hsd1.ma.comcast.net] entered the room. (01:29:30) base2design: mjl: how do you feel about the performance of ssl on inferno? (01:30:38) base2design: i've been playing around with your hget code (thank you VERY much) and at least on my macbook pro w/inferno, something is pretty slow. i'm trying to zone in on the slow area, but wanted to know if your experience matches mine. (01:30:54) base2design: (off to read about profiling inferno code...) (01:31:42) base2design: (it doesn't appear to be my server, as my crummy coldfire https client (non-inferno) is pretty zippy at the same thing) (01:36:44) mjl-: base2design: i don't really have an idea about it. (01:37:02) mjl-: i don't run an ssl server (it's on the todo list somewhere) (01:37:29) mjl-: what kind of slowness are you seeing? low throughput? or high cpu/mem usage while doing ssl? (01:37:38) mjl-: or slow connection set up? (02:55:01) The account has disconnected and you are no longer in this chat. You will be automatically rejoined in the chat when the account reconnects.