| 00:00.00 | landley | Didn't get into work until 11 today, and have to make up time from yesterday. |
| 00:01.13 | prpplague | ho ho hum |
| 00:02.00 | landley | I'm going to try to get the Impact userspace onto the nail, and when I do that I can try to build tinycc for an arm host, and get it to produce arm target binaries. |
| 00:02.05 | landley | I expect much to break, but I can fix it... |
| 00:03.31 | *** join/#edev prpplague^2 (n=dave123_@ppp-70-244-160-94.dsl.rcsntx.swbell.net) |
| 00:03.37 | prpplague^2 | jeeze |
| 00:04.55 | ds2 | landley: tinycc uses gas right? |
| 00:05.19 | CosmicPenguin | thats not carbon friendly |
| 00:06.08 | ds2 | CosmicPenguin: how slow is the Geode LX800 @ 533MHz compared to the other processors? |
| 00:06.18 | CosmicPenguin | pick a processor |
| 00:06.47 | ds2 | Turion 64 at 1.536GHz |
| 00:07.03 | ds2 | thinking of replacing my ailing Turion 64 with a Pepper Pad which has the Geode |
| 00:08.02 | CosmicPenguin | The LX800 probably has about half the performace |
| 00:08.03 | CosmicPenguin | give or take |
| 00:08.15 | landley | ds2: nope. |
| 00:08.18 | landley | It's completely standalone. |
| 00:08.27 | landley | Built-in assembler. (Currently only for x86.) |
| 00:08.38 | landley | If you don't provide assembly input, it outputs _machine_code_. |
| 00:08.41 | landley | Bytes. |
| 00:08.42 | ds2 | that's not bad... so the LX800 should be comparable to the Turion 64 in power save (800MHz) mode? |
| 00:08.59 | ds2 | landley: so it is not very portable then? |
| 00:09.03 | landley | ds2: geode vs intel atom. :) |
| 00:09.11 | landley | ds2: you have to write a new back-end for each target. |
| 00:09.24 | landley | Host sinde, it should be portable. |
| 00:09.31 | landley | In practice, it needs some debugging, but I'm working on that. |
| 00:09.35 | ds2 | landley: 2 problems there - 1. no sub $800 UMPC sized units, 2. I haven't seen one that would do NTSC out |
| 00:10.04 | CosmicPenguin | ds2: yes, give or take |
| 00:11.22 | CosmicPenguin | ds2: on the plus side, the Geode is fanless |
| 00:11.28 | ds2 | CosmicPenguin: is the AMD CS5536 buggy like the AMD 8111? (in your experience) |
| 00:11.48 | CosmicPenguin | completely different beast |
| 00:12.06 | ds2 | CosmicPenguin: yes, but I heard that upgrading the system to 1G of DDR causes heat issues on the unit that requires a clocking down to 500MHz from 533MHz |
| 00:12.23 | CosmicPenguin | On the pepper pad specifically? |
| 00:12.27 | CosmicPenguin | wow - bad EE - no cookie |
| 00:12.56 | ds2 | I see. The AMD 8111 left a very bad taste in my mouth (spent 2 months going back and forth with different people on an errata) |
| 00:13.02 | ds2 | yes, on the pepperpad |
| 00:14.04 | CosmicPenguin | well, errata happen |
| 00:14.33 | CosmicPenguin | but if the engineers are dicks about it, then its not cool |
| 00:14.37 | ds2 | yes, errata happens. I just wish it would be acknowledge a little better |
| 00:15.19 | ds2 | I was integrating a stock Opteron board and the mfg of the board couldn't get heads or tails from AMD until I start pressing them to inquire about a specific errata |
| 00:16.11 | ds2 | CosmicPengiun: so in practical apps, there shouldn't be a reason why I can't play DVD type VOBS on a LX800? |
| 00:16.28 | *** join/#edev prpplague^2 (n=dave123_@ppp-70-244-160-94.dsl.rcsntx.swbell.net) |
| 00:16.32 | CosmicPenguin | probably not DVD quality |
| 00:16.56 | ds2 | hmmm |
| 00:17.11 | CosmicPenguin | depends on the situation of course |
| 00:17.19 | *** mode/#edev [+o prpplague] by ChanServ |
| 00:17.40 | ds2 | the Turion64 I have seems to do okay even when I lock it down on 800MHz |
| 00:17.51 | CosmicPenguin | it probably does math better then we do |
| 00:17.58 | CosmicPenguin | are you deCSSing? |
| 00:18.22 | ds2 | no, nonecrypted VOB |
| 00:18.56 | CosmicPenguin | okay - if you use Xv you should be able to get 2 or 3 Mb |
| 00:19.10 | CosmicPenguin | especially if you have a smart integer based decoder |
| 00:19.46 | ds2 | I guess that is tolerable |
| 00:20.00 | CosmicPenguin | The Xv will do the YUV color conversion for you and save you a ton of CPU time |
| 00:20.14 | ds2 | would you have any feel on how the LX800 is compared to the transmeta stuff? |
| 00:20.31 | CosmicPenguin | well, seeing as though you can actually buy a LX.. :) |
| 00:20.36 | ds2 | I thought the graphics engine on the Geode can do YUV conversions? |
| 00:20.39 | CosmicPenguin | yep |
| 00:20.54 | ds2 | CosmicPenguin: i can get a transmeta system (used) |
| 00:21.00 | CosmicPenguin | I have to go - I'll be online later |
| 00:21.29 | prpplague | ho ho hum |
| 00:21.44 | prpplague | ~lart Mirell |
| 00:21.44 | purl | explains, ever so gently, that if Mirell doesn't give the channel more information, they can't help |
| 00:22.10 | prpplague | ~lart Mirell |
| 00:22.10 | purl | shoves a crumpet down Mirell's throat, happy now?! Huh? Want some JAM with that? |
| 00:34.17 | landley | What? |
| 00:34.22 | landley | (Has he even been on today?) |
| 00:36.03 | prpplague | landley LAK is driving me nuts |
| 00:37.11 | prpplague | ~seen landley |
| 00:37.14 | purl | landley is currently on #edev (6h 52m 27s) #uclibc (6h 52m 27s) #elinux (6h 52m 27s). Has said a total of 67 messages. Is idling for 2m 52s, last said: '(Has he even been on today?)'. |
| 00:37.52 | prpplague | landley hehe, just the slow methods of patch acceptance |
| 00:38.00 | prpplague | landley hows work? |
| 00:38.23 | landley | slow methods of patch acceptance? |
| 00:38.41 | prpplague | landley ben dooks is the sub arch maintainer for s3c24xx |
| 00:38.49 | landley | I can't "me too" the patch on the arm list. I haven't been subscribed for months and their web archive interface is deeply stupid. |
| 00:39.06 | prpplague | landley and he seems to very slow to accept patches that don't come from him or his company |
| 00:39.14 | landley | If you cc: me on a submission, I can acked-by it. |
| 00:39.30 | landley | Eh, I'm pretty good about going around that sort of thing. I'm just not in the loop this time. |
| 00:39.44 | landley | Give it a full week before reposting. |
| 00:39.45 | prpplague | landley RMK won't accept any s3c24xx patches without being signed off by bjdooks |
| 00:40.07 | prpplague | landley yea |
| 00:40.16 | landley | yes, but we can create a paper-trail of non-responsiveness in light of the upcoming -rc1 merge window and the obviously trivial nature of the patch. |
| 00:40.22 | prpplague | landley remember the BUG() issue i talked about in the past? |
| 00:40.33 | landley | This isn't really an s3c24xx patch. This is a new platform file. |
| 00:40.39 | landley | prpplague: um, no? |
| 00:40.51 | prpplague | landley but its a s3c24xx based devices, which has to go thru bjdooks |
| 00:41.17 | landley | Then we make him look very obviously like an unresponsive roadblock. |
| 00:41.28 | landley | You're the hardware engineer for the device, we have a patch that works. |
| 00:41.35 | prpplague | landley where in several places BUG() isn't defined so gcc complains about end of control structure |
| 00:41.49 | landley | Hadn't heard about it, but ok? |
| 00:42.06 | landley | I should resubscribe to the arm list, but I never read it unless somebody pointed out a thread. |
| 00:42.17 | landley | If they didn't have a BROKEN web interface, I'd be able to use that... |
| 00:42.32 | prpplague | landley http://www.arm.linux.org.uk/developer/patches/viewpatch.php?id=4562/1 |
| 00:43.17 | landley | ok |
| 00:43.22 | prpplague | landley we've discussed in the past as far back as 2.6.16 |
| 00:43.30 | landley | prpplague: got you beat. |
| 00:43.52 | prpplague | landley but get this, 2.6.25-rc as not one but 3 need examples of this issue |
| 00:43.55 | landley | http://127.0.0.1/hg/firmware/file/cc673b708b3b/sources/patches/linux-2.6.13-cmdline.patch |
| 00:44.04 | landley | Er, put "landley.net" in place of 127.0.0.1. |
| 00:44.15 | prpplague | hehe |
| 00:44.15 | landley | (Yes, I run a copy of my website on my laptop. Who doesn't?) |
| 00:44.35 | landley | prpplague: resubmit. |
| 00:44.50 | landley | cc: the arm list, but also cc: linux-kernel, and cc: andrew morton personally. |
| 00:44.58 | prpplague | landley yea i keep a copy of my web site as well: www.caribeanhotties.com |
| 00:45.07 | prpplague | jokes with landley |
| 00:45.11 | landley | Getting stuff into the -mm tree doesn't commit the upstream maintainer to take it, but it shames them into at least _responding_. |
| 00:46.51 | prpplague | i just can't understand why they don't want to fix it |
| 00:47.37 | prpplague | i know its not a complete build failure of the build, but darn, if you are going to have the option to define BUG() and nothing, you should have to code with that in mind |
| 00:47.55 | prpplague | s/and/as |
| 00:48.37 | ds2 | just the typical assert problem where people never assume assert can be turned off |
| 00:53.21 | landley | I'm installing slackware from dvd. |
| 00:53.34 | landley | Admittedly doing this under qemu from an iso file to an image file, but still. |
| 00:53.46 | landley | Somebody reported a bug with the impact build. :) |
| 00:55.12 | landley | It's something to do whihle reading about Long Salvo Recording. |
| 01:15.02 | *** join/#edev RobotGuy (n=robotguy@dsl093-038-072.pdx1.dsl.speakeasy.net) |
| 01:24.05 | *** join/#edev Sienna (n=Sienna@pool-71-255-208-156.bltmmd.east.verizon.net) |
| 01:24.53 | *** join/#edev chouimat|away (n=dieu@r2351064.cidc.net) |
| 01:44.32 | *** join/#edev dijenerate (n=dijenera@69.73.220.145) |
| 01:45.39 | ds2 | wow there are still slackware users?! |
| 01:51.50 | Redhatter | Yep |
| 01:52.18 | Sienna | mayhap they are just slacking moving to a new OS? |
| 01:52.26 | Sienna | (ok, my puns suck, this is normal :P) |
| 02:05.29 | *** join/#edev Appleman1234 (n=Appleman@131.181.47.37) |
| 02:10.17 | Redhatter | Sienna, If it ain't broke, why fix it?\ |
| 02:12.06 | Sienna | um... cause I like fixing working things :D |
| 02:34.46 | *** join/#edev CP|home (n=nobody@www.crouse-house.com) |
| 03:28.39 | *** join/#edev chouimat|away (n=dieu@r2351064.cidc.net) |
| 03:49.31 | *** join/#edev chouimat|away (n=dieu@r2351064.cidc.net) |
| 03:53.03 | *** join/#edev _ccjoe_ (n=ccjoe@75-169-37-101.slkc.qwest.net) |
| 04:00.16 | *** join/#edev bsdfox__ (n=h36sa@c-71-197-73-68.hsd1.ca.comcast.net) |
| 04:01.25 | *** join/#edev DrZimmerman (n=theo@tbaumgartner.broker.freenet6.net) |
| 05:32.37 | *** join/#edev harpal (n=harpal@121.246.75.165) |
| 07:24.47 | *** join/#edev [AD]Turbo (n=turbo@host131-128-dynamic.57-82-r.retail.telecomitalia.it) |
| 07:25.16 | [AD]Turbo | hi there |
| 07:42.36 | *** join/#edev linac (n=lin@121.41.36.186) |
| 08:14.56 | *** join/#edev lyakh (n=lyakh@p57BD2C3E.dip0.t-ipconnect.de) |
| 09:29.12 | *** join/#edev Appleman1234 (n=Appleman@131.181.47.9) |
| 09:55.36 | *** join/#edev Appleman1234 (n=Appleman@131.181.47.29) |
| 10:31.32 | *** join/#edev hw_ (n=hw@p578b3905.dip0.t-ipconnect.de) |
| 10:41.08 | *** join/#edev RobotGuy (n=robotguy@dsl093-038-072.pdx1.dsl.speakeasy.net) |
| 11:02.53 | *** join/#edev linac (n=lin@121.41.8.239) |
| 11:06.11 | *** join/#edev bbradley (n=bbradley@82-32-35-42.cable.ubr05.azte.blueyonder.co.uk) |
| 13:00.37 | *** join/#edev chouimat|work (n=dieu@209.217.106.98) |
| 13:17.23 | *** join/#edev GPSFan (n=kenm@12.10.255.246) |
| 13:17.23 | *** mode/#edev [+o GPSFan] by ChanServ |
| 13:17.39 | *** join/#edev prpplague (n=dave@mail.americanmicrosystems.com) |
| 13:22.41 | *** join/#edev sjhill (n=sjhill@real.realitydiluted.com) |
| 13:22.41 | *** mode/#edev [+o sjhill] by ChanServ |
| 13:22.47 | prpplague | sjhill: greetings |
| 13:22.56 | *** mode/#edev [+o prpplague] by ChanServ |
| 13:23.06 | chouimat|work | morning |
| 13:24.55 | prpplague | chouimat|work: greetings |
| 13:28.37 | prpplague | ho ho hum |
| 13:28.42 | prpplague | yet another fun day |
| 13:58.09 | sjhill | hello |
| 14:07.30 | chouimat|work | hmmm nice LPC3200 ---> ARM926EJ |
| 14:22.29 | *** join/#edev dijenerate (n=dijenera@66.54.118.16) |
| 14:26.02 | *** join/#edev emil|k (i=emil@e179235248.adsl.alicedsl.de) |
| 14:28.52 | *** join/#edev Sascha (n=sisawess@cssun.rrze.uni-erlangen.de) |
| 14:30.08 | *** join/#edev ce_geek (n=ce_geek@roger-blanchette-4.umelst.maine.edu) |
| 14:30.52 | ce_geek | prpplague: I'm still alive BTW. :) Thesis goes away on Friday (finally.) Two weeks until defense. Prep for that will be cake, so next week I'll be hammering lots. |
| 14:31.09 | prpplague | ce_geek: good deal |
| 14:31.28 | prpplague | ce_geek: i've been battling windmills so i'm totally frustrated and worn out |
| 14:31.38 | prpplague | ce_geek: so it was a good thing that you were busy |
| 14:32.50 | ce_geek | prpplague; Windmills? |
| 14:33.14 | ce_geek | prpplague: heh, I almost got thrown off my bicycle today by the wind on the way into work. It's a good day to own a windmill up here in Maine. |
| 14:34.30 | prpplague | was making a reference to "Don Quixote" |
| 14:35.16 | chouimat|work | prpplague: this is too obscure for most people ;) |
| 14:35.28 | chouimat|work | <PROTECTED> |
| 14:36.44 | prpplague | chouimat|work: the wikipedia entry or my reference to don quixote ? |
| 14:37.01 | chouimat|work | the don quixote reference :) |
| 14:38.46 | prpplague | ahh |
| 14:40.39 | prpplague | chouimat|work: did you get the reference? |
| 14:40.58 | chouimat|work | yes they forced me to read the book :( |
| 14:41.35 | prpplague | hehe |
| 14:48.41 | chouimat|work | prpplague: http://www.ietf.org/rfc/rfc5241.txt?number=5241 |
| 14:53.53 | prpplague | work safe? |
| 14:54.56 | chouimat|work | yes |
| 14:58.33 | ce_geek | prpplague: I wondered if I was just missing something :) |
| 14:58.39 | prpplague | hehe |
| 15:01.46 | ce_geek | Arright back to making citations and other miscellaneous corrections. Talk to you folks soon. |
| 15:02.13 | prpplague | ce_geek: later |
| 15:18.20 | *** join/#edev ico2 (n=ico2@modem-3087.chameleon.dialup.pol.co.uk) |
| 15:27.24 | *** join/#edev magnetron (n=magnetro@unaffiliated/magnetron) |
| 16:15.38 | prpplague | OMG this has got to be the best quote i've ever heard from microsoft - "ISO is too slow, explicit, and factual. We have several projects in the pipeline that are stalled because ISO likes to give people time to think. This is unacceptable." |
| 16:21.33 | agaffney | hah |
| 16:21.43 | SpeedEvil | ... |
| 16:24.01 | prpplague | that is sooooo weird, it's just too funny |
| 16:32.11 | *** join/#edev ico2 (n=ico2@modem-3087.chameleon.dialup.pol.co.uk) |
| 16:37.16 | *** join/#edev CosmicPenguin (n=nobody@12.190.142.202) |
| 16:37.17 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 16:50.24 | *** join/#edev CosmicPenguin (n=nobody@12.190.142.202) |
| 16:50.24 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 16:50.37 | *** join/#edev gandhijee (n=root@host-66-202-34-165.spr.choiceone.net) |
| 16:59.44 | prpplague | CosmicPenguin: re |
| 17:00.52 | CosmicPenguin | yo |
| 17:01.12 | prpplague | CosmicPenguin: whats cookin? |
| 17:01.52 | CosmicPenguin | nothing - just fighting bad wireless |
| 17:01.52 | prpplague | CosmicPenguin: price for lefthand brew is getting as silly as the gas prices :( |
| 17:01.55 | CosmicPenguin | they are probably related... :) |
| 17:02.03 | prpplague | CosmicPenguin: right now its $29.99 for a six pack |
| 17:02.09 | *** join/#edev bbradley (n=bbradley@82-32-35-42.cable.ubr05.azte.blueyonder.co.uk) |
| 17:08.28 | sjhill | speaking of drinks |
| 17:08.38 | sjhill | is it possible to buy Cocaine anymore? |
| 17:08.46 | sjhill | never got to try that liquid stuff |
| 17:11.46 | SpeedEvil | As your friendly neighbourhood retailer. |
| 17:11.56 | chouimat|work | :) |
| 17:13.45 | sjhill | oh, the website's still there |
| 17:13.53 | sjhill | dare i, or dare i not |
| 17:14.02 | sjhill | not sure what it would do to my training diet |
| 17:15.49 | *** join/#edev steliosk (n=Stelios@athedsl-389166.home.otenet.gr) |
| 17:27.41 | *** join/#edev CosmicPenguin (n=nobody@12.190.142.202) |
| 17:27.41 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 17:31.02 | *** join/#edev cajaka (n=none@cpe-24-58-193-200.twcny.res.rr.com) |
| 18:13.54 | *** join/#edev drath (i=vmaster@p5B07E028.dip.t-dialin.net) |
| 18:16.08 | *** join/#edev ali_as (n=as@ambix.plus.com) |
| 18:22.48 | *** join/#edev davidc__ (n=davidc__@S01060002b360aacd.vc.shawcable.net) |
| 18:29.31 | *** join/#edev ico2 (n=ico2@modem-3087.chameleon.dialup.pol.co.uk) |
| 18:33.58 | *** join/#edev RobotGuy (n=robotguy@dsl093-038-072.pdx1.dsl.speakeasy.net) |
| 18:34.19 | prpplague | RobotGuy: greetings |
| 18:34.58 | RobotGuy | HI |
| 18:35.37 | prpplague | RobotGuy: you get your jffs2 working? |
| 18:42.39 | chouimat|work | likes it when a new kernel packages removes over 25 bugs from his list :) |
| 18:45.49 | RobotGuy | No |
| 18:46.15 | prpplague | RobotGuy: you get my email about turning on mtd? |
| 18:52.33 | RobotGuy | Yeah, did that. Got to go now |
| 19:08.28 | *** join/#edev roxfan2 (n=dunno@4.219-200-80.adsl-dyn.isp.belgacom.be) |
| 19:14.33 | *** join/#edev CP|home (n=nobody@www.crouse-house.com) [NETSPLIT VICTIM] |
| 19:14.33 | *** join/#edev wmat (n=btraynor@gromit.mixdown.ca) [NETSPLIT VICTIM] |
| 19:15.45 | prpplague | CP|home: uh home? |
| 19:17.08 | prpplague | CP|home: boss lady ok? |
| 19:19.00 | chouimat|work | prpplague: http://tnx.nl/php.jpg |
| 19:19.08 | chouimat|work | it's worksafe |
| 19:19.14 | prpplague | looks |
| 19:19.37 | prpplague | hehe |
| 19:21.50 | prpplague | "the new Intel parts are priced initially between $45 and $160 in 1,000-unit quantities. " |
| 19:21.52 | prpplague | wowowow |
| 19:22.53 | chouimat|work | which parts? |
| 19:28.56 | prpplague | intel atom |
| 19:29.26 | agaffney | chouimat|work: hah |
| 19:41.03 | prpplague | thats an outrageous price |
| 20:16.15 | gandhijee | we have a menlo board at my office right now |
| 20:17.22 | gandhijee | we heard 100 bux max for CPU + chipset |
| 20:29.26 | *** join/#edev ico2_ (n=ico2@modem-589.chimpanzee.dialup.pol.co.uk) |
| 20:34.21 | prpplague | gandhijee: yikes |
| 20:34.42 | prpplague | gandhijee: that is just way outside the range i would have expected |
| 20:36.36 | *** join/#edev CosmicPenguin (n=nobody@12.190.142.202) |
| 20:36.36 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 20:37.20 | gandhijee | we'll see, hopefully the price will be much less that that |
| 20:37.22 | gandhijee | *than that |
| 20:38.55 | gandhijee | hey prpplague you know if the openocd svn site is down |
| 20:38.56 | gandhijee | ? |
| 20:38.57 | prpplague | gandhijee: its gonna have to be a pretty darn good chipset to compete with sub $15 ARM SoC's |
| 20:39.07 | prpplague | gandhijee: sorry i do not know |
| 20:39.18 | prpplague | gandhijee: i've not done an svn pull in a couple of weeks |
| 20:39.24 | gandhijee | O |
| 20:39.42 | gandhijee | the menlo dev platform already has all the cellular stuff on it |
| 20:39.48 | gandhijee | like the old PXA dev platforms |
| 20:40.47 | prpplague | i don't if i should scream or laugh |
| 20:41.16 | prpplague | just got the schematics for one of the handheld device cradles |
| 20:41.43 | prpplague | no usb enable control, no usb +5v detect, no control resistors |
| 20:42.01 | prpplague | and the usb ground is shorted to the shield ground |
| 20:42.55 | gandhijee | i bet alot of cradles are like that |
| 20:43.01 | prpplague | uh no |
| 20:43.16 | prpplague | gandhijee: only ones that you don't want to work |
| 20:43.32 | gandhijee | i did mean the cheaper china ones |
| 20:44.17 | *** join/#edev roxfan (n=dunno@142.62-136-217.adsl-dyn.isp.belgacom.be) |
| 20:45.19 | prpplague | gandhijee: well, even cheaper ones from china have to work |
| 20:45.29 | prpplague | gandhijee: as is, this won't work |
| 20:46.01 | prpplague | gandhijee: the usb driver expects to be told when a usb cable is attached |
| 20:46.18 | prpplague | gandhijee: if it can't detect +5v it can't turn on the usb port |
| 20:46.24 | gandhijee | O |
| 20:46.33 | prpplague | gandhijee: which in this case, it can even turn on the port since there is no enable circuit |
| 20:46.58 | gandhijee | please excuse my retardedness |
| 20:47.07 | prpplague | gandhijee: and even if the driver could turn on the usb port, it wouldn't work properly due to missing resistors and noisy grounding |
| 20:48.05 | GPSFan | prpplague: do you guys have design reviews? that sort of crap would never pass muster where I've worked in the past. |
| 20:48.30 | prpplague | GPSFan: we used to when we had an engineering manager |
| 20:48.41 | *** join/#edev ali_as_ (n=as@ambix.plus.com) |
| 20:48.45 | prpplague | GPSFan: i just went to the president of the company over this |
| 20:48.57 | prpplague | GPSFan: we are suppose to have a design review regarding this tomorrow |
| 20:49.04 | prpplague | GPSFan: its a horrific mess |
| 20:49.16 | GPSFan | prpplague: maybe you need to take your boss out for a beer, ;>) |
| 20:49.24 | CosmicPenguin | well - if its only at the schematic level its not so bad |
| 20:49.27 | CosmicPenguin | thats what reviews are for |
| 20:49.32 | CosmicPenguin | cluebats a'plenty |
| 20:50.52 | ali_as_ | Ahh, Hi Cosmic. I'm looking for x86 JTAG information, for example athlon, geode. prpplague has suggested you might be able to help. |
| 20:51.15 | GPSFan | prpplague: make sure you bring along the "reference design" doc's for your hub chip that shows how the enble ckt shoule be implemented. |
| 20:51.27 | gandhijee | ali_as_: you know if the openocd svn site is down? |
| 20:51.36 | chouimat|work | prpplague: another design by your favorite EE? |
| 20:53.37 | ali_as_ | gandhijee, the web version of the svn server is working fine. I don't have an svn client to try. |
| 20:54.34 | gandhijee | k |
| 20:54.56 | ali_as_ | Cosmic, preferentially debug through JTAG interformation. |
| 20:55.14 | CosmicPenguin | Um - okay |
| 20:55.18 | CosmicPenguin | what sort of information? |
| 20:56.30 | ali_as_ | Like how to debug at the lowest level. I've implimented debug protocols on ST20 from scratch and MIPS64 EJTAG systems almost from scratch. |
| 20:57.04 | roxfan | isn't all jtag pretty much standard? |
| 20:57.26 | roxfan | you send commands to switch states and read/write registers |
| 20:57.46 | ali_as_ | I have an old athlon, I don't have a jtag pinout, I understand IR is usually long on x86 systems 20bits or more so the chances of my being able to work it out myself are fairly slim. |
| 20:58.33 | ali_as_ | It's a bit like saying a computer is simple becuase logic 1 and logic 0 states are well documented :) |
| 20:58.54 | CosmicPenguin | yes, you would need published IR values |
| 21:01.40 | ali_as_ | I've been unable to find anything except a pin out for an 'athlon' jtag connector. |
| 21:02.10 | ali_as_ | I understand the new intel stuff can be debugged through the jtag port, but again, no information. |
| 21:02.12 | roxfan | http://nsa.unaligned.org/jrev.php try this maybe |
| 21:03.21 | CosmicPenguin | I'm not sure if that information is published or not |
| 21:03.24 | ali_as_ | That apears to be an interesting program, but it's doubtless boundary scan based. |
| 21:03.31 | CosmicPenguin | I'm pretty sure we don't publish it for the Geode |
| 21:04.31 | ali_as_ | That surprises me slightly given the almost embedded nature of the chip. |
| 21:04.44 | ali_as_ | Do you work for AMD then? |
| 21:05.56 | CosmicPenguin | I do |
| 21:06.54 | ali_as_ | That explains prpplague's suggestion! |
| 21:07.28 | ds2 | jtag and x86? heh |
| 21:07.40 | ali_as_ | macraigor's OCD commander claims to work with athlon chips, so I was hoping for a spec, either open or under nda. |
| 21:08.27 | CosmicPenguin | ds2: All of our hardware debuggers are JTAG based - its a sane interface |
| 21:09.01 | ali_as_ | And since Geode is targeted for embedded applications and the spec sheets make some mention of debug through jtag, I thought there might be tools/information or source code. |
| 21:09.14 | ds2 | I wonder how many people actually make their own x86 boards plus complete bootloader instead of buying an off the shelf board and using a stock bios |
| 21:11.11 | ds2 | and unless you are doing your own bios, is there any other good reason to use a low level debugger like jtag? |
| 21:11.25 | CosmicPenguin | ds2: very few - I only know of one commercial vendor that does that |
| 21:11.25 | *** join/#edev landley (n=landley@adsl-68-93-84-18.dsl.hstntx.swbell.net) |
| 21:11.25 | *** mode/#edev [+o landley] by ChanServ |
| 21:11.44 | ali_as_ | Might make sense for developing hardware. |
| 21:11.46 | CosmicPenguin | ds2: and no - JTAG isn't very interesting once you get past basic kernel init |
| 21:12.17 | landley | CosmicPenguin: doesn't it let you attach gdb to the hardware, the way qemu does? |
| 21:12.33 | CosmicPenguin | oh, heavens no |
| 21:12.50 | ds2 | CosmicPenguin: I'd think the onboard debugger should work for kernel init... it is the cold start code that would benefit the most from JTAG (BIOS) |
| 21:12.53 | landley | You can't hook gdb up to hammer via jtag? |
| 21:13.12 | ds2 | landley: You can. |
| 21:13.18 | landley | is confused. |
| 21:13.40 | ds2 | I suspect the offerings available for the Geode is a lot like the ICE olden days |
| 21:14.03 | CosmicPenguin | ds2: yeah - it is also slightly useful for integrated function development - its sometimes a lot easier to bang on the graphics engine with JTAG |
| 21:14.10 | ali_as_ | Cosmic, btw is the name a reference to 'Cosmic penguin' the music group? I knew someone at university that used to play their records. |
| 21:14.21 | ds2 | bound out processors and ice was a novel thing |
| 21:14.37 | CosmicPenguin | ali_as_: no, it is not |
| 21:14.41 | ali_as_ | Very strange electronic music. |
| 21:14.51 | ali_as_ | Fair enough. |
| 21:15.49 | roxfan | ali_as_: if ocd commander works, you can just look at how it does it |
| 21:16.00 | *** join/#edev dijenerate (n=dijenera@66.54.118.16) |
| 21:16.09 | ali_as_ | Thats part of the plan if I can't get any other information. |
| 21:16.13 | ds2 | AFAIK, the OCD stuff is for talking to the debugger in the ARM core |
| 21:16.37 | ds2 | and if so, using the OCD stuff is like using the BDM debuggers on an ARM |
| 21:16.43 | ali_as_ | Athlon though is very much on the way out, so it would be of limited usefulness. |
| 21:17.05 | CosmicPenguin | ali_as_: unfortunately, if the information isn't published, then I am of little use. I don't know if we have published the information, but if it is published, then it probably lives somewhere on amd.com |
| 21:17.40 | ds2 | ali_as_: the internal debugger is sufficient for most things |
| 21:17.44 | ali_as_ | I've googled, and can't find anything at all other than references to the capabilities in the spec sheets. |
| 21:18.11 | ds2 | be nice if they made an updated SoftICE for Linux though |
| 21:18.24 | roxfan | softice is dead |
| 21:18.28 | roxfan | if you didn't hear |
| 21:18.34 | ds2 | :) |
| 21:18.55 | ali_as_ | Unfortunatly that defeats my point. I'm making a hardware interface (isn't everyone) and I'm putting at much information in one directory on my hard disk so I can work out the best method to use. |
| 21:19.30 | ali_as_ | Softice will live on in our hearts and out memories ;D |
| 21:19.48 | prpplague | returns and reads the log |
| 21:20.06 | landley | prpplague: how big an ext2.gz can I feed the apex bootloader? |
| 21:20.32 | prpplague | landley: with the correct settings a compressed image of 8mb |
| 21:21.06 | prpplague | landley: really if you need bigger than that, best is to do a very small initrd and chroot to a jffs2 rootfs |
| 21:21.12 | prpplague | landley: or even boot straight to jffs2 |
| 21:21.42 | landley | prpplague: yeah, I just wanted to feed in my existing armv4l ext2 as a starting point. Oh well. |
| 21:22.08 | prpplague | landley: your existing armv4l initrd compressed is bigger than 8mb? |
| 21:22.30 | landley | It contains gcc and binutils. :) |
| 21:22.33 | prpplague | landley: btw, once you get get pass the linux kernel, its much easier to do gdb via rs-232 or network |
| 21:22.37 | prpplague | landley: oh |
| 21:22.45 | prpplague | landley: how big is the uncompress image? |
| 21:22.51 | landley | ~35 megs. |
| 21:23.12 | prpplague | landley: ahh, wouldn't have worked anyway, as the hammer on has 32mb of ram |
| 21:23.14 | landley | Compresses down to 13. |
| 21:23.45 | landley | For some reason I thought it had more than that... |
| 21:23.53 | landley | Haven't looked up the specs in a while. |
| 21:23.53 | *** join/#edev Sienna (n=Sienna@pool-71-246-73-161.bltmmd.east.verizon.net) |
| 21:23.55 | prpplague | landley: 32mb ram, and 16mb flash |
| 21:24.08 | landley | yup, I remember the ratio. |
| 21:24.37 | prpplague | landley: trying to think of the best combination for you |
| 21:25.16 | landley | I can plug in my external usb key or something. |
| 21:25.31 | landley | What I'm really trying to do is get tinycc on there. |
| 21:25.33 | prpplague | landley: yea , i was just going to ask if all your custom stuff is under /usr |
| 21:25.49 | landley | Currently everything's under /tools. |
| 21:25.53 | prpplague | landley: ahh |
| 21:25.58 | prpplague | landley: then what i'd do |
| 21:26.02 | landley | I know. |
| 21:26.14 | ds2 | Hmmm there is a modern day free softice thing that can work in Linux |
| 21:26.17 | prpplague | landley: is setup with a jffs2 root, and have all of the tools on your thumb drive |
| 21:26.17 | landley | Doesn't help booting in the first place. |
| 21:26.18 | ds2 | http://rr0d.droids-corp.org/ |
| 21:26.38 | landley | I was trying to avoid spending more than the ~3 minutes at a time I can justify as a distraction from work. |
| 21:26.40 | prpplague | landley: and then mount /tools from the thumb drive |
| 21:26.47 | prpplague | landley: hehe |
| 21:26.56 | landley | Making a new root file system image definitely gets into the "wait until after work" category. |
| 21:27.05 | landley | But lately, I've just been too tired. C++ is _exhausting_. |
| 21:27.17 | ds2 | C++ is a waste of time ;) |
| 21:27.26 | prpplague | landley: you got the image up on landley.net ? |
| 21:27.29 | ds2 | ducks while all the crap flies around ;) |
| 21:28.09 | Sienna | drops a piano on ds2 for fun :D |
| 21:28.31 | ali_as_ | C++ is a waste of CPU time. |
| 21:28.35 | prpplague | Sienna: you the same Sienna thats been posting on tincantools forum? |
| 21:28.38 | ali_as_ | hides in a piano proof shelter. |
| 21:28.40 | Sienna | yep |
| 21:28.45 | prpplague | ahh |
| 21:28.49 | ds2 | hey, you are suppose to use Anvils! |
| 21:29.04 | Sienna | I use pianos! I am classier! |
| 21:30.19 | ali_as_ | ds2, Never tried it, but I'm told windbg is the 'new softice'. |
| 21:30.49 | ds2 | easier solution - write bug free code :P |
| 21:32.57 | prpplague | Sienna: do you have a hammer board? |
| 21:33.06 | Sienna | not yet :( |
| 21:33.24 | Sienna | if I have money left over after I buy a 6dof, i iwll be getting one |
| 21:34.08 | prpplague | ahh |
| 21:34.21 | prpplague | Sienna: mind if i ask what your impressions are of hammer? |
| 21:34.29 | Sienna | sure, have at! |
| 21:34.51 | Sienna | (and thats why i am interested in the Hammer RDP btw, cause I want to put it on my bot) |
| 21:35.19 | Sienna | what did you want to know prpplague |
| 21:36.01 | prpplague | Sienna: just you initial impressions, what attracted you to the hammer, what concerns you about purchasing one, what kind of questions did you have that haven't been answered by the website |
| 21:36.50 | Sienna | initial impressions: Much denser then a pc104 board in terms of MIPS/sq in (which is good) |
| 21:37.17 | Sienna | Not as fast as a pc104 or pico itx (which for some apps is bad) |
| 21:37.35 | Sienna | however, i feel that the tradeoff is generally in favor of the Hammer |
| 21:38.21 | Sienna | Concerns: Well, what most people have. Small company, so is it going to be around? What kind of extensions are they coming out with? |
| 21:38.44 | Sienna | 5V to 3.3V interfacing is my biggest concern, due to my stock in 5/6V stuff atm |
| 21:39.40 | landley | prpplague: what image? |
| 21:39.42 | Sienna | Embedded linux worries me a bit, because it still seems like there is a lot of hacking to get buildroots and whatnot built. (a 'bootloader' esq function would maybe help me) |
| 21:39.53 | prpplague | landley: your 13mb test image |
| 21:40.02 | Sienna | (Background: I am a MAC girl, and I like things that "just work") |
| 21:40.43 | prpplague | Sienna: ahh interesting points |
| 21:40.49 | landley | prplague: prpplague: http://landley.net/code/firmware/downloads/image/qemu-image-armv4l.tar.bz2 |
| 21:40.49 | CosmicPenguin | there are things that "just work"? |
| 21:40.52 | landley | That's a touch out of date. |
| 21:41.04 | Sienna | um,.... I dunno what else. I am going to get one and hope I can get it to work and hope I can interface it, and hope I can put some IK code on it :D |
| 21:41.04 | landley | Sienna: c++ _is_ a waste of time, but I'm getting paid to do it until the end of July. |
| 21:42.22 | Sienna | hmm... netsplit |
| 21:42.22 | prpplague | Sienna: i'll see if we can post some info to address some of those concerns |
| 21:42.22 | prpplague | Sienna: tincantools is actually a sister company of a rather big company, http://www.amltd.com |
| 21:42.22 | Sienna | i guess my biggest hurdles in robots is ... electronics. |
| 21:42.22 | prpplague | Sienna: so its going to be around for awhile |
| 21:42.35 | prpplague | Sienna: that seems to be very common |
| 21:42.35 | Sienna | I am very good mechanically, and decent for SW (considering thats my BS) |
| 21:42.45 | prpplague | Sienna: yea, that seems to be a big issue in the robotics area |
| 21:42.47 | ds2 | buildroot is not the only way of doing things |
| 21:43.41 | *** join/#edev GPSFan (n=kenm@12.10.255.246) [NETSPLIT VICTIM] |
| 21:43.41 | *** join/#edev dijenerate (n=dijenera@66.54.118.16) [NETSPLIT VICTIM] |
| 21:43.41 | *** join/#edev ali_as_ (n=as@ambix.plus.com) [NETSPLIT VICTIM] |
| 21:43.41 | *** join/#edev gandhijee (n=root@host-66-202-34-165.spr.choiceone.net) [NETSPLIT VICTIM] |
| 21:43.41 | *** join/#edev Sascha (n=sisawess@cssun.rrze.uni-erlangen.de) [NETSPLIT VICTIM] |
| 21:43.41 | *** join/#edev samuelr (n=samuelr@dsl093-061-094.pit1.dsl.speakeasy.net) [NETSPLIT VICTIM] |
| 21:43.41 | *** join/#edev MonMotha (i=monmotha@c-69-245-153-193.hsd1.in.comcast.net) [NETSPLIT VICTIM] |
| 21:43.41 | *** join/#edev SpeedEvil (n=mauve@tor/regular/SpeedEvil) [NETSPLIT VICTIM] |
| 21:43.41 | *** join/#edev lSpeedEvi (n=mauve@212.159.7.170) [NETSPLIT VICTIM] |
| 21:43.41 | *** mode/#edev [+o GPSFan] by irc.freenode.net |
| 21:44.10 | prpplague | jokes with ds2 |
| 21:44.10 | Sienna | Hello again netsplit people! |
| 21:44.10 | prpplague | ds2: says who? imho buildroot is the only way |
| 21:44.10 | ds2 | :P |
| 21:44.24 | CosmicPenguin | but you are clearly crazy |
| 21:44.24 | Sienna | prpplague, what do they do, it looks like inventory control? |
| 21:44.24 | landley | does not use buildroot. |
| 21:44.28 | ali_as_ | Hi from the least popular side of the split. |
| 21:44.56 | Sienna | landley, well, you must think you are a god or something :D |
| 21:45.10 | landley | Sienna: no, I just wrote my old cross compiling build system. |
| 21:45.17 | landley | http://landley.net/code/firmware |
| 21:45.29 | landley | Waiting for 2.6.25 to do another release... |
| 21:45.31 | ds2 | there should be no need for a cross build "system" :P |
| 21:45.53 | landley | ds2: ok, I scripted the process of building a minimal native build environment so I can then build natively. |
| 21:46.05 | landley | Or at least under qemu. |
| 21:46.09 | Sienna | let me put it this way... get it down where all I have to do is write C, type 'make', and have the file sent serial to the Hammer, and I would be happy! |
| 21:46.12 | landley | Because doing it by hand (ala cross linux from scratch) is a royal pain. |
| 21:46.20 | *** join/#edev roxfan2 (n=dunno@142.62-136-217.adsl-dyn.isp.belgacom.be) |
| 21:46.22 | landley | Sienna: I'm working on that. |
| 21:46.33 | ds2 | landley: I put the blame for that squarely on the autoconf folks |
| 21:46.41 | landley | Right now you can type "./build.sh armv4l" and it'll create a cross compiler, kernel, and ext2 image. |
| 21:46.50 | landley | Alas, the ext2 image won't fit on hammer and the kernel has the wrong .config. |
| 21:46.53 | landley | Working on it. :) |
| 21:46.54 | ds2 | make CC=arm_flavorOftheee-gcc all |
| 21:46.58 | ds2 | that should be all you need |
| 21:46.59 | Sienna | cool! Then all I need else is for TCT to make the RDP so that it has that 5V interface board ! |
| 21:48.01 | Sienna | why should i need to build a kernel? The hardware is the same for all Hammers |
| 21:48.13 | ds2 | software features |
| 21:48.42 | landley | Sienna: alas, the kernel .config controls a lot more than just what hardware it runs on. |
| 21:48.55 | Sienna | I am looking for a "Hammer.h" that has functions like "Send_I2C(string tosend)" and "Output(int port, int data)" |
| 21:48.57 | landley | What drivers are built, what filesystems are built in... |
| 21:49.43 | landley | The default hammer kernel it comes with doesn't have a driver for most of the external USB devices I want to plug into it, for one thing. |
| 21:49.58 | Sienna | oh, I would be ok with that :D |
| 21:50.23 | Sienna | but that is me... I am looking for a fast I/O board, with some limited math functions |
| 21:50.34 | Sienna | I don't want to plug in a keyboard and wifi and such |
| 21:50.47 | SpeedEvil | math functions would be faster onboards |
| 21:51.07 | landley | SpeedEvil: hammer hasn't got a floating point coprocessor. |
| 21:51.12 | SpeedEvil | yes. |
| 21:51.14 | landley | I'm using libc level soft-float. |
| 21:51.15 | Sienna | How many matrix multiplies can a hammer do? |
| 21:51.19 | SpeedEvil | But USB1.1 is _SLOW_ |
| 21:51.20 | landley | (You could also do kernel level soft float, but that's slower.) |
| 21:51.38 | SpeedEvil | unless things are abnormal, it'll be faster onboard |
| 21:51.56 | Sienna | again, personal goal, solving IK is what I would hope to get done on board |
| 21:52.10 | SpeedEvil | ick? |
| 21:53.10 | Sienna | Inverse Kinimatics... e.g., knowing the starting point of a body, figure out the angles of linked appendages such that the leg ends up in a certain x,y,z point in space |
| 21:53.12 | prpplague | returns from phone call |
| 21:53.33 | Sienna | its a lot of linear algebra |
| 21:53.45 | SpeedEvil | Well... |
| 21:54.03 | Sienna | and i think another term too... but I forget. |
| 21:54.04 | prpplague | Sienna: yea industrial data collection devices |
| 21:54.16 | SpeedEvil | Unless you have more than a few dozen linkages in series, I wouldn't have thought that it had a point. |
| 21:54.27 | Sienna | Basically, its the entire focus of things like Calc 3, all that matrix stuff |
| 21:54.57 | prpplague | Sienna: interesting info |
| 21:54.57 | ali_as_ | Any inertia included? |
| 21:55.05 | prpplague | makes a copy of the irc log |
| 21:55.11 | SpeedEvil | If you have 'just' something like 4 or 5 joints in series, I wouldn't have thought the maths are that heavy. |
| 21:55.22 | Sienna | SpeedEvil, its the fundamentals of moving around legged robots, or industrial arms, etc |
| 21:55.33 | SpeedEvil | Sienna: I'm not disagreeing with that. |
| 21:56.16 | SpeedEvil | Sienna: just that I debate - based on a few moments thought rather than research - that for normal apps, the hammer would be unable to cope with the maths. |
| 21:56.27 | prpplague | <Sienna> why should i need to build a kernel? The hardware is the same for all Hammers |
| 21:56.36 | SpeedEvil | Especially as most of it could be done fixed-point. |
| 21:56.38 | prpplague | Sienna: not everyone wants to do the same thing with their hammer board |
| 21:56.39 | ali_as_ | Fix point maths should be fine for that. |
| 21:56.58 | prpplague | Sienna: the number of possible configuration options is very large |
| 21:57.11 | Sienna | sec, phone |
| 21:57.15 | SpeedEvil | Sienna: consider that many people may not want to use |
| 21:57.45 | prpplague | Sienna: but for the, RDP we can definetly produce a kernel and rootfs to be very standardized for the type of usage you want |
| 21:57.49 | SpeedEvil | (for example) I2C. This will bloat the kernel by some K. The memory on the hammer is very constrained. |
| 21:59.25 | Sienna | prpplague: I know, I was just saying what I wanted to do with my hammer boards. I am not trying to generalize your entire customer base |
| 21:59.40 | prpplague | Sienna: right |
| 21:59.58 | Sienna | how much memory does a module loader take/ |
| 22:00.00 | Sienna | ? |
| 22:00.02 | prpplague | Sienna: what we'll probably do is have a kernel image and rootfs specifically for the RDP |
| 22:00.14 | prpplague | Sienna: it varies with the module |
| 22:00.20 | Sienna | it would be appreciated prpplague |
| 22:01.02 | Sienna | I was just trying to hypothesize whether a module loader, and prebuilt modules would be smaller then a all inclusive RDP kernel |
| 22:01.57 | prpplague | Sienna: ahh |
| 22:02.12 | prpplague | Sienna: yea, we'll probably have a combination of the two |
| 22:02.31 | Sienna | but yeah, you asked for my wish list, that was it :D |
| 22:02.51 | prpplague | Sienna: but i think from what i've heard from robotguy and you, we probably need to add a small api for you guys to do lowlevel access to the i2c,spi, and other ports |
| 22:02.57 | Sienna | do you have any estimated beta timeframe for the RDP? |
| 22:03.15 | prpplague | landley: not sure i understand the structure of that rootfs image |
| 22:03.28 | ali_as_ | Just looking at pico itx - wow. |
| 22:03.44 | prpplague | landley: what do you pass as the init= for the kernel? |
| 22:04.04 | prpplague | Sienna: probably 30-60 days |
| 22:05.21 | Sienna | can i be a beta tester! |
| 22:05.22 | prpplague | Sienna: i'm sure we will have 3 to 5 beta testers, and the boss has already noticed your input on the forum |
| 22:05.22 | Sienna | also, for development purposes, what happens if I feed the current carrier board 6V? |
| 22:05.34 | prpplague | Sienna: so i suspect you'll be on the list |
| 22:05.39 | Sienna | yay! |
| 22:06.20 | *** join/#edev Appleman1234 (n=Appleman@131.181.47.3) |
| 22:06.25 | Sienna | hello Appleman1234 |
| 22:07.07 | Appleman1234 | Hi |
| 22:08.26 | landley | prpplague: it's a bzipped ext3 file. |
| 22:08.42 | landley | I decompress it and then feed it to qemu as a hard drive image, and do root=/dev/hda |
| 22:08.48 | Sienna | prpplague, if I can suggest a name for the RDP, I suggest the "Screw". :D I saw a Nail somewhere, so that was taken :P |
| 22:08.53 | prpplague | Sienna: the current design is based on regulated input to the carrier board, however you could run at it at 6v just understand that the usb host +5v won;t be correct |
| 22:09.02 | landley | I was going to gzip it and feed it in as an initrd with rdinit=/tools/bin/bash or some such. But it won't fit. |
| 22:09.23 | ds2 | prpplague: will the LDO handle the extra heat? |
| 22:09.42 | ds2 | 5 to 6V is nearly 50% increase in heat onthe LDO |
| 22:09.43 | prpplague | landley: so in qemu the root looks like what? |
| 22:09.54 | prpplague | ds2: 6v is the tops |
| 22:10.07 | prpplague | ds2: and its low enough power that its not a problem |
| 22:10.09 | Sienna | If i need to feed it only 5V, i will find a way, its just the main power bus on my bot is 6, that is all |
| 22:10.31 | ds2 | prplague: okay... the gumstix folks recommend using no more then 4.5V in certain configs due to heat |
| 22:10.33 | landley | Hi Appleman1234. |
| 22:10.35 | prpplague | Sienna: yea, if you are planning to do this, i'd just get a small ldo and drop it to 5v |
| 22:10.46 | landley | prpplague: in qemu the root=/dev/hda |
| 22:11.00 | landley | It looks like an unpartitioned hard drive containing an ext2 image starting at offset 0. |
| 22:11.08 | prpplague | landley: sorry, and i meant was if you did a ls in / , what does it look like? |
| 22:11.19 | landley | It says "tools/" and nothing else. |
| 22:11.27 | landley | Old Linux From Scratch trick. |
| 22:11.38 | landley | The shell is /tools/bin/bash (or /tools/bin/ash) |
| 22:11.44 | Appleman1234 | Hi landley |
| 22:11.47 | Sienna | for some reason i have the old zelda (like zelda 1) song stuck in my head |
| 22:11.53 | Sienna | </random> |
| 22:12.05 | landley | Sienna: the mario bros theme will drive that out. |
| 22:12.16 | prpplague | landley: hmm |
| 22:12.19 | Sienna | I know zelda a lot more :D |
| 22:12.20 | prpplague | landley: that is interesting |
| 22:12.41 | landley | After work I can build a filesystem without development tools in it, my scripts have a switch for that. |
| 22:12.51 | landley | I've just got another hour or so before I can justify the time. :) |
| 22:13.28 | Sienna | excuses... not like you haven't been talking anyway |
| 22:13.32 | prpplague | landley: you actually have your image set for 248mb |
| 22:13.32 | landley | prpplague: if you download http://landley.net/code/firmware/downloads/image/qemu-image-arv4l.tar.bz2 |
| 22:13.41 | landley | You can extract that, and then run ./run-armv4l.sh |
| 22:13.51 | landley | It should load it into qemu for you, and boot to a shell prompt within your console window. |
| 22:14.02 | prpplague | i don't have qemu loaded |
| 22:14.16 | landley | prpplague: I've tweaked the image sizes to give me some spare space. The current ones mount a /dev/hdb for empty space. |
| 22:14.31 | prpplague | landley: ahh |
| 22:14.43 | landley | Sienna: worst case scenario, Wierd Al drives just about anything else out. |
| 22:15.03 | Sienna | :( Worst case for sure |
| 22:15.06 | landley | Basically I'm trying to get enough of a native development environment to rebuild itself under itself from source code. |
| 22:15.17 | landley | I have 6 more commands needed before I'm there. |
| 22:15.22 | Sienna | eep! The gentoo way! |
| 22:15.22 | landley | (Listed in host-tools.sh) |
| 22:15.38 | landley | Sienna: without the need for python or all the infighting on the list, sure. |
| 22:15.47 | landley | Without any package management at all, really. It's a bash script. |
| 22:15.53 | landley | Half the point is you can read the script and see what it's _doing_. |
| 22:16.16 | prpplague | landley: so when you are in qemu and have this image booted the only thing in / is the directory tools? |
| 22:16.18 | landley | prpplague: the six commands are: bzip2 find install od sort diff |
| 22:16.24 | landley | The busybox versions (if any), don't work right. :( |
| 22:16.32 | landley | prpplague: yup. |
| 22:16.36 | prpplague | interesting |
| 22:16.40 | landley | That's so you can build a new system in there without file collisions. |
| 22:16.46 | landley | As I said, old Linux From Scratch trick. |
| 22:16.53 | Sienna | very odd |
| 22:17.22 | landley | That's also why mini-native.sh has these lines near the beginning: |
| 22:17.25 | landley | export UCLIBC_DYNAMIC_LINKER=/tools/lib/ld-uClibc.so.0 |
| 22:17.26 | landley | export UCLIBC_RPATH=/tools/lib |
| 22:17.52 | landley | My gcc wrapper script feeds gcc -nostdlib -nostdinc and then rebuilds the paths from scratch, explicitly. |
| 22:17.59 | landley | Those environment variables tell it to use some nonstandard values... |
| 22:21.57 | prpplague | landley: you normally include the man and info pages ? |
| 22:23.47 | prpplague | landley: hmm, question |
| 22:24.14 | prpplague | landley: you know you have two copies of the compile tools like ar in the image? |
| 22:24.24 | *** join/#edev GPSFan (n=kenm@12.10.255.246) |
| 22:24.24 | *** join/#edev dijenerate (n=dijenera@66.54.118.16) |
| 22:24.24 | *** join/#edev ali_as_ (n=as@ambix.plus.com) |
| 22:24.24 | *** join/#edev gandhijee (n=root@host-66-202-34-165.spr.choiceone.net) |
| 22:24.24 | *** join/#edev Sascha (n=sisawess@cssun.rrze.uni-erlangen.de) |
| 22:24.26 | *** join/#edev samuelr (n=samuelr@dsl093-061-094.pit1.dsl.speakeasy.net) |
| 22:24.26 | *** join/#edev MonMotha (i=monmotha@c-69-245-153-193.hsd1.in.comcast.net) |
| 22:24.26 | *** join/#edev SpeedEvil (n=mauve@tor/regular/SpeedEvil) |
| 22:24.26 | *** join/#edev lSpeedEvi (n=mauve@212.159.7.170) |
| 22:24.26 | *** mode/#edev [+o GPSFan] by irc.freenode.net |
| 22:26.04 | prpplague | landley: and you have duplicate copies for c++/g++ and gcc/gcc-4.1.2 |
| 22:27.38 | landley | prpplague: not intentionally. (the man and info pages.) |
| 22:27.44 | landley | It may install them by default, I'm not trying to clean it out. |
| 22:27.49 | prpplague | landley: ahh ok |
| 22:27.52 | landley | It's gdb/binutils. Attempting to size optimize it is a waste of time. |
| 22:28.01 | landley | It's from the FSF. It is bloated. This is a tautology. |
| 22:28.20 | landley | (All syllogisms have three parts, therefore this is not a syllogism.) |
| 22:28.50 | landley | prpplague: actually, they're not quite duplicates. |
| 22:29.00 | landley | It has to do with the way gcc does stuff. |
| 22:29.05 | landley | One copy is wrappers that calls the other copy. |
| 22:29.09 | prpplague | landley: they have the exact same file size and crc |
| 22:29.19 | landley | In that case they're probably hard links. |
| 22:29.21 | landley | Check the link count. |
| 22:29.34 | landley | (It's quite possible I broke the link count making the ext2 image.) |
| 22:29.46 | landley | If they were identical, then they _were_ hardlinks. |
| 22:29.51 | landley | At one point. |
| 22:29.58 | prpplague | landley: yea, thats what i suspect |
| 22:30.28 | prpplague | landley: well heading home |
| 22:30.31 | prpplague | later folks |
| 22:30.32 | prpplague | bbiab |
| 22:30.33 | Sienna | night~! |
| 22:31.08 | landley | I was about to tell prpplague that ar is hardlinked here. :) |
| 22:31.23 | landley | I'll hae to check the ext2 image. I thought "tar" handled maintaining hardlinks. Possibly I need to hit it with a larger brick. |
| 22:31.50 | landley | Ah, it's user mode linux. |
| 22:32.07 | landley | UML is using hostfs, which probably doesn't propogate hard links from the parent filesystem. |
| 22:32.43 | landley | Ok, I need to create the tarball from the host system and just _extract_ it in the UMl script. Right... |
| 22:33.51 | landley | And the reason I was doing that is I needed to strip off the top level directory. Hmmm... |
| 22:44.19 | ds2 | don't see a good reason to shrink gcc... it is glibc that is the abomination |
| 22:48.00 | SpeedEvil | But how else do you get gcc onto the flash? |
| 22:48.46 | CosmicPenguin | you don't |
| 22:48.47 | CosmicPenguin | ever |
| 22:49.19 | landley | CosmicPenguin: you use tinycc instead. :) |
| 22:49.32 | CosmicPenguin | don't compile on target |
| 22:49.33 | CosmicPenguin | ever |
| 22:49.37 | landley | Me, I run gcc under qemu and have distcc dial out to the cross compiler so the build is merely horribly slow rather than insanely so. |
| 22:49.48 | landley | CosmicPenguin: native compiling is inherently easier than cross compiling. |
| 22:50.00 | CosmicPenguin | Thats not a valid excuse |
| 22:50.01 | landley | Muchly. |
| 22:50.09 | landley | Yes, it is. |
| 22:50.14 | CosmicPenguin | its easier because people are idiots |
| 22:50.17 | landley | There are about 600 packages that can be cross compiled if you really try. |
| 22:50.25 | landley | There are something like 30,000 in the Ubuntu repository. |
| 22:50.47 | landley | It's not "people are idiots", it's "tracking two contexts is an order of magnitude harder than tracking one". |
| 22:50.56 | landley | Gnu make is hard enough to get right at the best of times. |
| 22:51.02 | landley | Let alone ./configure. (*shudder*) |
| 22:51.23 | CosmicPenguin | I disagree - its not at all difficult to make your code cross compile friendly |
| 22:52.11 | Sienna | if he wants to compile natively, let him! |
| 22:52.40 | Sienna | don't come all high and mightly and say 'don't, it is verbotten!' :( |
| 22:52.43 | landley | CosmicPenguin: it's easy to make _my_ code cross compile friendly. |
| 22:52.52 | landley | It's a royal pain to make other people's code do so after the fact. |
| 22:53.09 | landley | Especially if they make extensive use of bad ideas like automake. |
| 22:53.44 | landley | And if you start saying "I will only ever use good code", you've ruled out about half of gcc. |
| 22:53.54 | CosmicPenguin | absolutely - it is a royal pain in the ass |
| 22:53.59 | landley | )The implementation of that thing is a crawling horror in places.) |
| 22:54.01 | CosmicPenguin | but its a fight worth fighting |
| 22:54.08 | landley | doesn't think so. |
| 22:54.17 | landley | Throwing hardware at the problem is easier than throwing engineering time at the problem. |
| 22:54.45 | landley | Computers are fast, emulators work, and with the distcc trick I'm within an order of magnitude of native compiling speed and dropping my constant factor as I come up with tweaks. |
| 22:55.01 | landley | s/native compiling speed/cross compiling speed/ |
| 22:55.05 | landley | purl: shut up. |
| 22:55.06 | purl | as you wish, master landley |
| 22:55.09 | CosmicPenguin | I far prefer having my quad opteron compile my Geode kernel in 5 minutes then having the Geode do it in 30 minutes |
| 22:55.25 | CosmicPenguin | and qemu is a cute trick, but its not a universal one |
| 22:55.32 | Crofton | but the geode does a better job ... |
| 22:55.47 | CosmicPenguin | Crofton: I didn't say quad core.... :D |
| 22:55.56 | landley | My laptop compiles my kernel in something like 2 minutes, and it's only 2 way. How many modules do you have enabled? |
| 22:56.15 | landley | CosmicPenguin: qemu is only a few years old, and adding new platforms all the time. |
| 22:56.24 | landley | I'd rather put my effort into qemu than playing whack a mole with 30k packages. |
| 22:56.32 | landley | And then re-fixing them every release. |
| 22:56.44 | landley | (Keep in mind "developers who cross compile" is < 1% of the developer base.) |
| 22:57.03 | landley | An awful lot of project maintainers think "native compiles on everything" is the end of the matter. |
| 22:57.09 | CosmicPenguin | Well, they are both engineering problems, just in different areas |
| 22:57.13 | landley | They actively don't WANT to complicate their build process with cross compiling stuff. |
| 22:57.35 | landley | And thus resist patches to that effect, because they can't test them and don't want to maintain them. |
| 22:57.41 | landley | Besides, getting perl to cross compile is just evil. |
| 22:57.50 | CosmicPenguin | It can be done and done well |
| 22:58.01 | landley | A friend of mine had to debug a perl x86->mips cross compile that was using x86 signal numbers on mips. |
| 22:58.09 | landley | LITTLE things break. Subtle things you notice after you've shipped. |
| 22:58.17 | landley | Doesn't happen nearly so much with native compiling. |
| 22:58.25 | landley | Because there arent two contexts to leak into each other. |
| 22:58.33 | landley | CosmicPenguin: deploying on windows can be done and done well. |
| 22:58.41 | landley | It's possible to get a turtle airborne by strapping enough rockets to it. |
| 22:58.43 | landley | Doesn't make it a good idea. |
| 22:58.58 | ali_as_ | goes to find a turtle. |
| 22:59.11 | landley | ali_as_: There might be a #logo channel here, dunno. |
| 22:59.37 | CosmicPenguin | but can it be cross-compiled? :) |
| 22:59.47 | ali_as_ | Why would I want to strap rockets to a logo? |
| 23:00.09 | ali_as_ | >:) |
| 23:00.31 | Sienna | logo!!! *memories* :D |
| 23:01.34 | ali_as_ | Ahh, logo a programming language for the turtle. |
| 23:01.46 | ali_as_ | I was thinking a real turtle. |
| 23:02.53 | ali_as_ | I will call him super turtle, master of the water, master of the land and now master of space. |
| 23:03.30 | ali_as_ | I may need a long plaque. |
| 23:03.40 | SpeedEvil | Naah. |
| 23:03.47 | SpeedEvil | Dymo labelmaker on the shell. |
| 23:04.13 | ali_as_ | Just been looking at the pico itx board, very powerful, but seems to have virtually no raw connectivity. |
| 23:06.38 | landley | wants to lobby the texas state legislature to install a historic plaque commemorating its own placement. |
| 23:06.46 | ds2 | the question there really is - is your quad opteron stable with lots of memory (just saying thanks for the errata about not being able to drive certain combinations of memory) |
| 23:06.49 | landley | "On this site, on [such and such a date], a historic marker was placed." |
| 23:06.57 | ali_as_ | I'd vote for that. |
| 23:07.02 | landley | lives in Austin, the capital of texas, where the legislators hang out anyway... |
| 23:07.45 | ali_as_ | Perhaps better still, a plaque commemerating the loss of a previous plaque. |
| 23:08.03 | landley | What I can't figure out is if I should draw up the proposal to have it installed _in_ the capitol building (inside the debating chamber, in front of the speaker's podium). |
| 23:08.07 | landley | Or put it at UT somewhere... |
| 23:08.16 | landley | The first would assume that they don't actually read the text of the bills they pass... |
| 23:08.42 | landley | ali_as_: but that isn't properly recursive. |
| 23:08.44 | ali_as_ | It should be stapled the the president and transferred every time a new one is elected. |
| 23:08.56 | landley | Nah, duct tape. |
| 23:09.21 | ali_as_ | I don't suffer from the linux virus that makes people demand acronyms be self referentially recursive. |
| 23:09.56 | landley | ali_as_ that's a gnu virus, not Linux. |
| 23:10.11 | ali_as_ | I stand corrected. |
| 23:10.12 | landley | And it's not an acronym, it's a historic marker. |
| 23:10.34 | ali_as_ | Meh, its bleed over from the same mnemes. |
| 23:10.59 | Appleman1234 | lol |
| 23:11.03 | landley | despises the word "meme". |
| 23:11.28 | Appleman1234 | I have a thecus n1200 it has a mpc8347 board with a powerquicc ppc processor :) |
| 23:11.40 | landley | good for you. |
| 23:11.43 | Appleman1234 | I am customising the linux on it :) |
| 23:12.03 | ali_as_ | All that means is you have a meme that doesn't like the word meme.... |
| 23:12.34 | Appleman1234 | the serial just gives out garbage so I must have the wrong settings |
| 23:13.44 | *** join/#edev prpplague (n=dave123_@ppp-70-244-160-94.dsl.rcsntx.swbell.net) |
| 23:13.53 | Sienna | wb prpplague |
| 23:14.00 | prpplague | thanks |
| 23:14.32 | landley | prpplague: you were right. |
| 23:14.35 | prpplague | ?? |
| 23:14.47 | landley | I was packaging up the image with User Mode Linux, via hostfs, and hosfts was breaking the symlinks. |
| 23:15.07 | landley | er, hard links. |
| 23:15.10 | prpplague | ahh |
| 23:17.20 | landley | I just checked in an updated package-mini-native.sh to fix it, but I won't upload new prebuilt images until I do a release. |
| 23:17.20 | landley | Which is after 2.6.25 comes out. |
| 23:17.20 | prpplague | np |
| 23:17.20 | landley | Thanks for the bug report. :) |
| 23:17.20 | prpplague | np |
| 23:17.20 | prpplague | landley one thing i also noticed |
| 23:17.20 | prpplague | landley in tools/bin |
| 23:17.21 | landley | What about it? |
| 23:17.21 | prpplague | landley ar didn't have a prefix |
| 23:17.22 | prpplague | landley all the other tools were prefixed but armv4l-unknown |
| 23:17.59 | landley | Hmmm, apparently nothing in binutils does. |
| 23:18.10 | landley | is looking. |
| 23:18.15 | landley | Did I upgrade binutils recently? |
| 23:18.46 | prpplague | not sure |
| 23:18.57 | landley | Rhetorical question. :) |
| 23:19.03 | prpplague | i know |
| 23:19.05 | prpplague | just being silly |
| 23:19.29 | landley | For some reason, I'm explicitly setting --program-prefix to the empty string in the build. |
| 23:19.38 | landley | Ah, hang on. |
| 23:19.46 | landley | _none_ of these tools should have a prefix. They're native tools. |
| 23:19.54 | prpplague | ahh |
| 23:19.55 | landley | So why do some of them, the breakage is at the other end... |
| 23:20.00 | landley | Why is gcc prefixing these... |
| 23:20.47 | landley | Ah, it's the c++ stuff. Which I've never properly tested... |
| 23:21.00 | prpplague | ahh ok, just wanted to make sure |
| 23:21.05 | landley | Well, it's all hardlinked anyway. |
| 23:21.22 | landley | (And actually, yes I _did_ test it. The C++ native build does indeed build some C++ stuff. I forget what, but it worked...) |
| 23:21.31 | landley | It's the cross compiler that doesn't have C++ support. |
| 23:21.47 | landley | Nor will building C++ take advantage of the distcc trick, but if your'e building c++ you already expect it to be insanely slow. |
| 23:22.01 | prpplague | yea |
| 23:22.23 | prpplague | how much did space did you save with the hardlinks corrected? |
| 23:22.35 | landley | Ok, so the four things with "armv4l-*" prefixes each have a hard link count of 3 or higher. |
| 23:22.49 | landley | They're alternate names for "gcc", "cc", "c++", and "g++". |
| 23:23.05 | landley | prpplague: about a meg and a half compressed. |
| 23:23.07 | landley | Good deal, that. |
| 23:23.10 | landley | Uncompressed, no idea. |
| 23:23.32 | prpplague | thats gonna be about 3meg uncompressed |
| 23:23.43 | landley | More, possibly. |
| 23:23.52 | landley | Redundant data compresses well, and if the files were adjacent in the tar... :) |
| 23:24.44 | landley | That said, I have an urge to delete "armv4l-linux-*" out of tools/bin from a sheer clutter perspective. |
| 23:24.58 | landley | It doesn't really take up a noticeable amount of extra space but I didn't ask gcc to do that. |
| 23:25.25 | prpplague | every mb counts |
| 23:25.55 | prpplague | landley boss lady needs some theraflu, gotta run to the grocery store, bbiaf |
| 23:26.14 | landley | ok |
| 23:26.30 | prpplague | sigwife ya know |
| 23:28.15 | landley | yup |
| 23:28.41 | Sienna | sigwife? |
| 23:29.49 | CosmicPenguin | its the higher priority equivelent of SIGGF |
| 23:30.09 | SpeedEvil | SIGWIFE is trapped though somewhat. |
| 23:30.29 | Sienna | :roll |
| 23:30.38 | Sienna | men and linux jokes :P |
| 23:31.45 | landley | I suspect Val Henson could probably make equivalent jokes in the other direction... |
| 23:32.22 | Sienna | val henson? |
| 23:32.24 | landley | And I know for a fact Admiral Hopper had plenty of hardware jokes. |
| 23:32.46 | Sienna | you are losing me |
| 23:32.48 | landley | http://valhenson.livejournal.com |
| 23:32.52 | landley | Female senior kernel developer. |
| 23:33.02 | Sienna | nice |
| 23:33.07 | landley | Oh there are several. |
| 23:33.13 | SpeedEvil | Admiral Grace Hopper. |
| 23:33.20 | landley | Pat Gaughen is short for "Patricia". |
| 23:33.25 | SpeedEvil | She was a pioneer in the computing field. |
| 23:33.30 | landley | One of the indian ones whose name I can never remember... |
| 23:33.43 | landley | Grace hopper _invented_the_compiler_. |
| 23:34.00 | landley | It doesn't get any more pioneer than that in software, with the possible exception of Ada Lovelace. |
| 23:34.03 | landley | (Who invented _software_.) |
| 23:34.10 | landley | Well, the first programmer. |
| 23:34.14 | Sienna | funny, i don't remember learning that.... of course, i don't remember much history at all, let alone compiler history |
| 23:34.20 | landley | She never got to actually run anything, but that was clearly a hardware problem. |
| 23:34.29 | landley | (Darn Charles Babbage with the feature creep...) |
| 23:35.03 | landley | http://www.sdsc.edu/ScienceWomen/hopper.html |
| 23:35.23 | landley | Her keynote speech at the History of Programming Languages conference was MARVELOUS. |
| 23:35.39 | landley | I was like 4 years old when it came out, but the University of Texas Library has a copy of the proceedings. |
| 23:36.20 | landley | GOOD GRIEF! |
| 23:36.21 | landley | http://www.youtube.com/watch?v=57bfxsiVTd4 |
| 23:36.33 | landley | I had no idea that footage existed. |
| 23:37.44 | Sienna | neat! |
| 23:42.15 | landley | I had no idea Grace Hopper had been such a marvelous guest on David Letterman's show... |
| 23:43.15 | Sienna | "when the general asks why it takes so damn long to communicate via satilite, you pull out these wires and tell him there are many nanoseconds between here and the satillite" :D |
| 23:43.48 | landley | :) |
| 23:44.00 | Sienna | (i work for a defense contractor... that is probably funnier to me then it should be) |
| 23:47.20 | *** join/#edev steliosk (n=Stelios@athedsl-4399403.home.otenet.gr) |
| 23:47.28 | landley | This is good too: http://www.youtube.com/watch?v=7sUT7gFQEsY |
| 23:47.46 | prpplague | returns |
| 23:48.04 | landley | prpplague: there's video of grace hopper on youtube! |
| 23:48.06 | landley | Who knew? |
| 23:48.44 | *** join/#edev ico2 (n=ico2@modem-589.chimpanzee.dialup.pol.co.uk) |
| 23:49.15 | prpplague | landley hehe |
| 23:50.28 | landley | I'd always heard about her handing out nanoseconds, but they have it on video... |
| 23:50.33 | prpplague | landley ok so its going to be high noon (actually 10am) at the engineering lab tomorrow |
| 23:50.47 | landley | prpplague: cool. |
| 23:50.52 | prpplague | bookmarks the youtube urls |
| 23:51.02 | prpplague | landley not really, gonna be nasty |
| 23:51.39 | landley | Good to get it _resolved_. |
| 23:52.10 | Sienna | whats getting resolved? |
| 23:52.26 | prpplague | Sienna got a problem with a lead EE at AML |
| 23:52.29 | landley | Sienna: prpplague's company is being destroyed by an idiot hardware engineer. |
| 23:52.33 | prpplague | Sienna long story |
| 23:52.37 | Sienna | :( |
| 23:52.43 | SpeedEvil | Have you considered trial by combat? |
| 23:52.53 | prpplague | Sienna note - aml not tincantools |
| 23:53.00 | prpplague | SpeedEvil indeed |
| 23:53.05 | Sienna | i saw :D |
| 23:53.14 | landley | Basically a series of bad technical decisions that are adding up and derailing project ship schedules. |
| 23:53.16 | Sienna | I vote feed him to a tape library..... |
| 23:53.57 | prpplague | well we will see how tomorrow goes |
| 23:54.19 | prpplague | i suspect tomorrow will only be a stop-gap |
| 23:54.24 | landley | "You don't have an ounce of nostalgia in you, do you?" "No." |
| 23:54.41 | Sienna | ugh... i hate the "not invented here" syndrome |
| 23:54.42 | landley | (Watching the 60 minutes interview with Grace Hopper.) |
| 23:55.02 | Sienna | (where "here" is the desk of the boss promoted past his / her competence) |
| 23:55.25 | Sienna | oh, I thought it applied to prpplague's situation... did seem to fit ;D |
| 23:55.52 | landley | alas, the boss in this case was promoted from marketing, from what I understand. |
| 23:56.03 | Sienna | evil! |
| 23:56.14 | landley | shrugs. |
| 23:56.18 | landley | I actually understand marketing. |
| 23:56.23 | landley | (I can't do sales, but I can do marketing.) |
| 23:56.37 | landley | (Look: shiny thing! Here's why it's shiny! Isn't shiny thing great?) |
| 23:56.44 | landley | Actually deciding to buy it is your problem, not mine. |
| 23:56.47 | prpplague | yea promoted from sales department |
| 23:56.48 | ds2 | has anyone evaluated marketing relative to a magic 8 ball? |
| 23:57.13 | ds2 | i.e. over a reasonably period, do any marekting dept beat or at least match a magic 8 ball ? |
| 23:57.51 | Sienna | but marketing is evil.... "look shiny happy thing! Gives you the moon on a string! Ready in only three months!" and when the engineers hear what marketing has promised, they look bad when it is only "swiss cheese from the moon, delivered in 12 months: |
| 23:58.10 | landley | Lying about the product is evil, sure. |
| 23:58.31 | landley | keep in mind that "Linux advocacy" is a marketing task. |
| 23:59.05 | Sienna | then you also have to fight program managers, who think that anytime an engineer gives an estimated time to complete, it is really only half of what they said |
| 23:59.40 | ds2 | simple solution there... increase all estimates by a factor of 4 ;) |
| 23:59.52 | Sienna | but then you would be a bad engineer by lying |
