highaltitude.log.20100807

[00:08] MoALTz (~no@188.146.200.233.nat.umts.dynamic.eranet.pl) left irc: Quit: Leaving
[00:10] SpeedEvil (~user@tor/regular/SpeedEvil) left irc: Read error: Operation timed out
[00:21] <Randomskk> anyone know what the ascent rate model is for balloons?
[00:21] <Randomskk> I could do the maths I guess
[00:31] <Randomskk> obviously constant. duh.
[01:11] SpeedEvil (~user@tor/regular/SpeedEvil) joined #highaltitude.
[01:25] LazyLeopard (~irc-clien@chocky.demon.co.uk) left irc: Quit: Bye
[01:39] SpeedEvil1 (~user@tor/regular/SpeedEvil) joined #highaltitude.
[01:40] SpeedEvil (~user@tor/regular/SpeedEvil) left irc: Ping timeout: 240 seconds
[01:42] Nick change: SpeedEvil1 -> SpeedEvil
[01:50] SpeedEvil (~user@tor/regular/SpeedEvil) left irc: Quit: Leaving.
[01:50] SpeedEvil (1000@tor/regular/SpeedEvil) joined #highaltitude.
[01:56] jasonb (~jasonb@dsl092-009-225.sfo1.dsl.speakeasy.net) left irc: Ping timeout: 260 seconds
[01:59] juxta (juxta@ppp203-122-193-94.static.internode.on.net) joined #highaltitude.
[02:20] ContraSF (email@89.180.152.111) joined #highaltitude.
[02:22] juxta (juxta@ppp203-122-193-94.static.internode.on.net) left irc: Ping timeout: 258 seconds
[02:25] DaveyC (~IceChat7@188-221-51-13.zone12.bethere.co.uk) left irc: Quit: If at first you don't succeed, skydiving is not for you
[02:54] jasonb (~jasonb@m3e0536d0.tmodns.net) joined #highaltitude.
[03:00] jasonb (~jasonb@m3e0536d0.tmodns.net) left irc: Ping timeout: 240 seconds
[03:30] jasonb (~jasonb@adsl-66-124-73-250.dsl.sntc01.pacbell.net) joined #highaltitude.
[05:32] shauno (~soneil@pdpc/supporter/professional/shauno) joined #highaltitude.
[07:24] natrium42 (~natrium@CPE000625d867e2-CM0014045885be.cpe.net.cable.rogers.com) joined #highaltitude.
[07:24] LA3QMA1 (~kgb@125.119.189.109.customer.cdi.no) joined #highaltitude.
[07:27] LA3QMA (~kgb@69.119.189.109.customer.cdi.no) left irc: Ping timeout: 240 seconds
[07:37] LA3QMA (~kgb@69.119.189.109.customer.cdi.no) joined #highaltitude.
[07:38] LA3QMA1 (~kgb@125.119.189.109.customer.cdi.no) left irc: Ping timeout: 260 seconds
[07:52] SpikeUK (~chatzilla@host81-151-38-237.range81-151.btcentralplus.com) joined #highaltitude.
[08:05] Upu (~Upu@ubn.upuaut.net) joined #highaltitude.
[08:13] TraumaPony (~TraumaPon@124-171-195-55.dyn.iinet.net.au) left irc: Ping timeout: 240 seconds
[08:13] natrium42 (~natrium@CPE000625d867e2-CM0014045885be.cpe.net.cable.rogers.com) left irc: Quit: Leaving
[08:14] TraumaPwny (~TraumaPon@203-214-123-83.dyn.iinet.net.au) joined #highaltitude.
[08:15] Nick change: TraumaPwny -> TraumaPony
[08:22] TraumaPwny (~TraumaPon@203-214-123-83.dyn.iinet.net.au) joined #highaltitude.
[08:26] TraumaPony (~TraumaPon@203-214-123-83.dyn.iinet.net.au) left irc: Ping timeout: 276 seconds
[08:31] Nick change: TraumaPwny -> TraumaPony
[08:58] MoALTz (~no@188.147.194.187.nat.umts.dynamic.eranet.pl) joined #highaltitude.
[09:50] MindSpigot (~random@78-86-167-25.zone2.bethere.co.uk) left irc: Read error: Connection reset by peer
[09:51] MindSpigot (~random@78-86-167-25.zone2.bethere.co.uk) joined #highaltitude.
[09:55] juxta (juxta@ppp203-122-193-94.static.internode.on.net) joined #highaltitude.
[10:05] <fsphil> morning one and all
[10:06] earthshine_ (~earthshin@212.183.140.33) joined #highaltitude.
[10:06] <Randomskk> yo
[10:06] earthshine_ (earthshin@212.183.140.33) left #highaltitude.
[10:28] <DanielRichman> morning
[10:29] <Randomskk> yo
[10:29] <Randomskk> by the way DanielRichman I also installed unattended-upgrades so in theory if you leave upgrades for a bit it should autoinstall them
[10:29] <DanielRichman> OK
[10:32] <Randomskk> I also started hacking on a python predictor for the heck of it
[10:32] <Randomskk> I think it could work
[10:32] <Randomskk> GFS tiles are half a degree each way
[10:32] <Randomskk> so requesting each one really shouldn't be all that bad
[10:34] <Randomskk> sadly the GFS data is indexed by latitude and longitude, so my earlier plan of using meters in a cartesian system and converting back at the end (makes applying wind speeds very, very easy) isn't that practical
[10:34] <Randomskk> happily it is also fairly easy to find a new lat,lon from an old lat,lon, a distance and a bearing
[10:34] <Randomskk> and converting the u,v wind velocities to polar is easy
[10:34] earthshine_2 (~earthshin@212.183.140.35) joined #highaltitude.
[10:35] earthshine_2 (earthshin@212.183.140.35) left #highaltitude.
[10:35] <Randomskk> I may see if I can use meters and only convert to lat, lon when crossing a tile boundary
[10:35] <DanielRichman> hmm 'cause I'm also more convinced that I can tidy the C one up
[10:35] <DanielRichman> can you mix printf and puts calls?
[10:35] <Randomskk> it does look like it needs it
[10:36] <Randomskk> I'm mostly doing this as a vague proof of concept, I don't imagine it'l be as fast as the C one
[10:36] <Randomskk> you /can/
[10:36] <Randomskk> I wouldn't
[10:37] earthshine_2 (~earthshin@212.183.140.35) joined #highaltitude.
[10:37] earthshine_2 (earthshin@212.183.140.35) left #highaltitude.
[10:52] <DanielRichman> Do you have an example scenario ini file to hand?
[10:53] <Randomskk> yea
[10:53] <DanielRichman> Cool, can I have it?
[10:53] <Randomskk> http://pastebin.com/JU6uF2CV
[10:53] <DanielRichman> ty
[10:54] <Randomskk> p.s. having a scenario.ini is a weird weird way of doing it
[10:54] <Randomskk> we'd be so much better served by yaml
[10:54] <Randomskk> also I don't like the data file format, it's just a load of random numbers on lines, what
[10:54] <Randomskk> drop the INI parser, get a YAML parser and use it for scenario and wind data tbh
[10:55] <DanielRichman> radical changes
[10:56] <DanielRichman> yaml is nice though, I agree. I'll go and search for a YAML C parser
[10:56] <Randomskk> yea but inis are silly and the wind data file is almost worse
[10:56] <DanielRichman> oh there's even libyaml in the repos
[10:58] <Randomskk> perfect.
[10:58] <Randomskk> omg yayayay
[10:58] <Randomskk> Dear MR Adam Greig
[10:58] <Randomskk> The following products relating to your Farnell order 94648354 have now been shipped from our warehouse.
[10:58] <Randomskk> 72-8230 DIGITAL STORAGE SCOPE, 60MHZ, 2 CH
[10:59] <Randomskk> http://uk.farnell.com/jsp/search/productdetail.jsp?CMP=i-ddd7-00001003&sku=1739444 :o
[11:01] <DanielRichman> farnell are very fast
[11:04] <DanielRichman> hmm looks like libyaml gives ye data SAX style
[11:04] <DanielRichman> the ini library rjw had was just iniparser_get_double("options:winderror") forex
[11:04] <DanielRichman> obviously less efficient but easy as hell to use
[11:07] <jonsowman> morning kiddies
[11:07] <jonsowman> Randomskk: just finishing coffee then jumping in car
[11:10] <jonsowman> did I hear the words python predictor? :D
[11:12] <DanielRichman> yeah he's finally snapped. Too much time working on php, I think
[11:13] <jonsowman> haha
[11:21] <Randomskk> :D
[11:21] <Randomskk> back
[11:23] LazyLeopard (~irc-clien@chocky.demon.co.uk) joined #highaltitude.
[11:24] RocketBoy (~steverand@217.47.75.8) joined #highaltitude.
[11:24] <jonsowman> Randomskk: anything I should bring?
[11:25] <jonsowman> i'll assume not
[11:25] <jonsowman> see you in a bit
[11:26] <Randomskk> can't think of anything
[11:26] <Randomskk> laptop?
[11:29] <DanielRichman> Randomskk: want to agree on a YAML format?
[11:29] <DanielRichman> eg http://pastebin.com/viX3fgUB
[11:30] MoALTz (~no@188.147.194.187.nat.umts.dynamic.eranet.pl) left irc: Quit: Leaving
[11:31] <Randomskk> do we need atmosphere
[11:31] <Randomskk> really
[11:32] <Randomskk> anyway that's fine, yes
[11:33] <DanielRichman> kool
[11:35] RocketBoy (~steverand@217.47.75.8) left irc: Quit: RocketBoy
[11:37] RocketBoy (~steverand@217.47.75.8) joined #highaltitude.
[11:39] <DanielRichman> Randomskk: I have to say though; at the moment implementing the YAML parser is looking like significantly more effort than the ini parser
[11:39] g8khw-iPhone (~RocketBoy@217.47.75.8) joined #highaltitude.
[11:40] g8khw-iPhone (RocketBoy@217.47.75.8) left #highaltitude.
[11:40] <Randomskk> hmm really?
[11:40] <Randomskk> libyaml might have some easy way to just filename->array or whatever
[11:40] <Randomskk> I haven't used it
[11:40] <DanielRichman> yeah that's what I was hoping for, but I don't think so
[11:40] <DanielRichman> initial_lat = iniparser_getdouble(scenario, "launch-site:latitude", 0.0);
[11:40] <Randomskk> yaml is deliciously easy in interpreted languages
[11:42] <Randomskk> data = yaml.load("data.yml"); lat = data['launch-site']['latitude']
[11:42] <Randomskk> delicious
[11:42] <DanielRichman> stop it.
[11:44] <Randomskk> =D
[11:45] <DanielRichman> I'm going to put YAML on the todo if that's ok with you
[11:46] <Randomskk> fair enough, there are better things to worry about atm
[11:50] <DanielRichman> hmm; going to go back on what I said earlier; how about these changes? http://pastebin.com/v72AV4jS
[11:52] <Randomskk> what's that?
[11:52] <Randomskk> INI? YAML?
[11:52] <Randomskk> looks like INI?
[11:52] <Randomskk> or is this more a proposed schema?
[11:53] <Randomskk> if the latter then better due to wind error being moved to gfs
[11:53] <DanielRichman> yeah that's a proposed new INI layout
[11:54] <DanielRichman> almost directly translatable to yaml; change the - to _; remove the []s and add a few :
[11:54] <Randomskk> works for me
[11:57] juxta (juxta@ppp203-122-193-94.static.internode.on.net) left irc: Ping timeout: 260 seconds
[11:58] <DanielRichman> oh and because I'm nuking most of the command line options I'm adding these (optional) entries: http://pastebin.com/9WAfnVkc
[11:58] <DanielRichman> although I'm keeping the cache-dir --flag
[12:02] <DanielRichman> http://pastebin.com/gU2C3kuA
[12:04] <Randomskk> shouldn't descending be altitude model
[12:05] <DanielRichman> I don't think so
[12:05] <Randomskk> eh
[12:05] <Randomskk> I guess you can make a good argument either way
[12:05] <DanielRichman> by that logic the launch-site:altitude should eb altitde-model:initial-altitude
[12:06] <Randomskk> true
[12:06] <DanielRichman> Randomskk: if I'm combing this code through, is there any advantage to swapping from float to double?
[12:06] <Randomskk> eeh
[12:06] <Randomskk> doubles take a little more CPU and RAM, but have higher resolution... on a modern computer the difference is usually negligable?
[12:06] <Randomskk> typedef it
[12:06] <Randomskk> but like
[12:07] <DanielRichman> well we'll be running this on a 64bit cpu now
[12:07] <Randomskk> true
[12:07] <Randomskk> typedef it anyway
[12:07] <Randomskk> and for the moment have it as a double
[12:08] <DanielRichman> what name do you suggest for the typedef?
[12:08] <Randomskk> what's it being used for?
[12:08] <Randomskk> something relevent to its use
[12:08] <Randomskk> typedef Altitude double
[12:08] <Randomskk> or maybe typedef Position double, then use it for alt+lat+lon
[12:08] <DanielRichman> Ok
[12:09] <DanielRichman> I'm going for prediction_val since it'll be used in alt lat lon various altitude model vars, wind speed etc
[12:10] <Randomskk> eeeh
[12:10] <Randomskk> so make multiple typedefs
[12:10] <Randomskk> you might want to make some things floats and some doubles
[12:10] <Randomskk> e.g. maybe windspeed is only given as a float accuracy anyway
[12:11] <DanielRichman> hmm ok. but lat and lon ought and alt ought to be on the same typedef since some functions work on 3d space etc.
[12:12] <Randomskk> yea
[12:15] RocketBoy (~steverand@217.47.75.8) left irc: Quit: RocketBoy
[12:15] DaveyC (~IceChat7@188-221-51-13.zone12.bethere.co.uk) joined #highaltitude.
[12:17] RocketBoy (~steverand@217.47.75.8) joined #highaltitude.
[12:50] RocketBoy (~steverand@217.47.75.8) left irc: Quit: RocketBoy
[13:17] <DanielRichman> jonsowman: do you know, are the GFS datasets always released exactly on time?
[13:30] jasonb (~jasonb@adsl-66-124-73-250.dsl.sntc01.pacbell.net) left irc: Ping timeout: 240 seconds
[13:31] Jasperw (~jasperw@212-98-44-230.static.adslpremium.ch) left irc: Ping timeout: 265 seconds
[13:31] jasonb (~jasonb@adsl-66-124-73-250.dsl.sntc01.pacbell.net) joined #highaltitude.
[13:34] Action: SpeedEvil lols at the concept of windspeed accurate to the limits of a float.
[13:40] LazyLeopard (~irc-clien@chocky.demon.co.uk) left irc: Quit: Bye
[13:45] Jasperw (~jasperw@212-98-44-230.static.adslpremium.ch) joined #highaltitude.
[13:49] <Randomskk> DanielRichman: jonsowman is here, but they are not, they vary a bit - you can look at past ones for an idea
[13:49] <DanielRichman> ok
[13:49] <Randomskk> by 'on time' what do you mean though
[13:49] <Randomskk> e.g. the 00z dataset is released about 0430 UTC
[13:49] <Randomskk> bbl soldering
[13:51] spenguin[work] (~penguin@59.162.86.164) left irc: Read error: Connection reset by peer
[13:51] spenguin[work] (~penguin@59.162.86.164) joined #highaltitude.
[14:00] spenguin[work] (~penguin@59.162.86.164) left irc: Read error: Connection reset by peer
[14:01] dcrand (~d_crand@68-117-98-7.dhcp.eucl.wi.charter.com) joined #highaltitude.
[14:03] spenguin[work] (~penguin@59.162.86.164) joined #highaltitude.
[14:05] <SpikeUK> DanielRichman many congrats for letter and pics in Wired mag
[14:24] SpikeUK (~chatzilla@host81-151-38-237.range81-151.btcentralplus.com) left irc: Quit: ChatZilla 0.9.86 [Firefox 3.5.11/20100701023340]
[14:27] LA3QMA (~kgb@69.119.189.109.customer.cdi.no) left irc: Read error: Operation timed out
[14:38] <DanielRichman> Randomskk: was thinking; and I reckon one of the major advantages of a pythonic or part-pythonic predictor would be ballast-drop-decision modelling. There's not yet any code for a balloon with a leak in it, and every time you reprogrammed the flight 'pooter you'd have to recompile the predictor, unless you fancy writing some sort of interpreted language for dropping ballast - or use python
[15:18] LA3QMA (~kgb@150.120.189.109.customer.cdi.no) joined #highaltitude.
[15:20] LA3QMA1 (~kgb@150.120.189.109.customer.cdi.no) joined #highaltitude.
[15:21] LA3QMA1 (~kgb@150.120.189.109.customer.cdi.no) left irc: Client Quit
[15:21] LA3QMA1 (~kgb@150.120.189.109.customer.cdi.no) joined #highaltitude.
[15:23] LA3QMA (~kgb@150.120.189.109.customer.cdi.no) left irc: Ping timeout: 265 seconds
[15:33] MoALTz (~no@188.146.115.2.nat.umts.dynamic.eranet.pl) joined #highaltitude.
[16:19] jcoxon (~jcoxon@host86-163-184-138.range86-163.btcentralplus.com) joined #highaltitude.
[16:32] <jcoxon> 2 balloons up in the US
[16:32] <jcoxon> on spacenear.us
[16:43] dcrand (d_crand@68-117-98-7.dhcp.eucl.wi.charter.com) left #highaltitude ("Leaving").
[16:44] <fsphil> ah brilliant
[16:46] <jcoxon> though only one is being properly tracked
[16:46] <jcoxon> the other is on aprs
[16:46] <fsphil> yea it's a bit spotty
[16:49] <fsphil> nice open landscape though, shouldn't be a problem retreating n9qgs-
[16:49] <jcoxon> just need a few more listeners
[16:55] <jcoxon> so much space...
[16:55] <sbasuita> no predictions though?
[16:55] <jcoxon> well no one updated the gfs files
[16:55] <jcoxon> it has to be done manually currently
[16:55] <sbasuita> add to feature list ;)
[17:00] <fsphil> there they go
[17:02] <fsphil> or not
[17:02] <jcoxon> looks like both are down - looking at spacenear.us and aprs.fi
[17:08] <DanielRichman> implement it yourself sbasuita >.> that'll make 3 of us hacknig the predictor at once
[17:14] Action: jcoxon 's fsa03 breakout boards have been shipped from seeedstudio - expect them in 3 months time then!
[17:19] <DanielRichman> does it really take that long?
[17:20] <jcoxon> hehe it varies
[17:20] <DanielRichman> >.<
[17:20] <jcoxon> i think Randomskk's board took a while
[17:20] <jcoxon> mine took about 2-3 weeks
[17:20] <jcoxon> last time i ordered
[17:20] <DanielRichman> I need to work out where those chips have got to; then order the alien2 ones asap
[17:21] <griffonbot> @BASE_DePauw: BASE 49 in corn Recovery in progress #ARHAB #GPSL [http://twitter.com/BASE_DePauw/status/20562604759]
[17:30] jcoxon (~jcoxon@host86-163-184-138.range86-163.btcentralplus.com) left irc: Quit: Leaving
[19:07] jcoxon (~jcoxon@host86-163-184-138.range86-163.btcentralplus.com) joined #highaltitude.
[19:08] <jcoxon> evening all
[19:29] <fsphil> hiya jcoxon, how's things today?
[19:30] <jcoxon> hey fsphil good thanks
[19:30] <jcoxon> hows you?
[19:31] <fsphil> not bad. full stomach, can't complain :)
[19:31] <jcoxon> hehe
[19:32] <jcoxon> just looking on the wiki - see that m1x10 has added a section about the fsa03 with on request data strings
[19:32] <jcoxon> that makes me very happy
[19:32] <jcoxon> instead of monitoring nmea
[19:35] <fsphil> yea it should make things a bit simpler
[19:35] <fsphil> though it requires a tx line, but I think most people have that
[19:40] Laurenceb (~laurence@host86-152-35-235.range86-152.btcentralplus.com) joined #highaltitude.
[19:40] <Laurenceb> hi
[19:40] <fsphil> hiya Laurenceb
[19:42] <DanielRichman> ping Upu
[19:44] <griffonbot> @BASE_DePauw: BASE 48 landed in corn east of Paris Recovered #ARHAB #GPSL [http://twitter.com/BASE_DePauw/status/20571191818]
[20:12] <Randomskk> back
[20:12] <Randomskk> yaaay
[20:12] <Randomskk> friggin TQFP chips though
[20:12] <Randomskk> the ARM Farnell sent me had a load of slightly bent inwards pins
[20:13] <Randomskk> jonsowman and I didn't notice how bad until it was too late
[20:13] <Randomskk> but it ended up with like, a couple of corner (bent) pins touching, and the rest being held above the board
[20:13] <Randomskk> so not only did almost none solder, but to solder the rest you had to bend them backwards, which makes them bridge and makes removing bridges a nightmare
[20:13] <Randomskk> we spent hours and hours carefully soldering each pin and trying to remove all the bridges
[20:13] <Randomskk> rest of the PCB went absolutely lovely
[20:13] <Randomskk> powered it up, no smoke, but it wouldn't program
[20:14] <Randomskk> messed with it for another couple of hours, no luck, jonsowman had to leave, I went over each pin again, prodding them with a needle to see if they moved, soldering them down, removing the bridges I made
[20:14] <Randomskk> tried again, nothing, and again, nothing, and then suddenly it worked
[20:14] <Randomskk> and now the ARM programs fine
[20:15] <jcoxon> hooray
[20:16] <Randomskk> lesson: check the pins are not bent before reflowing, ugh
[20:16] <Randomskk> but besides that yay
[20:16] <Laurenceb> what reflow oven did you use?
[20:17] <Randomskk> my reflow oven
[20:17] <Laurenceb> oh hot plate?
[20:17] <Randomskk> no, the grill
[20:17] <Laurenceb> ah :P
[20:17] <Randomskk> at uni it's an electric hob, but at home we have gas hobs, so I put it under the grill
[20:17] <Laurenceb> neat
[20:17] <Randomskk> there's a discussion to be had on top-down vs bottom-up heating I'm sure
[20:17] <Randomskk> but it works really quite well
[20:17] <Laurenceb> I've got the parts for makng a proper rflow oven with linear halogen IR lamps
[20:18] <Laurenceb> got the lamps as well... just need to build it and write some firmware
[20:18] <jcoxon> hey Laurenceb
[20:18] <jcoxon> long time
[20:18] <Laurenceb> Im going to try and interface a thermocouple directly with an attiny
[20:18] <Laurenceb> - been VERY busy lately
[20:19] <jcoxon> Laurenceb, whatch up to these days?
[20:19] <Laurenceb> in fact I havent slept since friday morning
[20:19] <jcoxon> oh right
[20:19] <jcoxon> that busy
[20:19] <Laurenceb> heh
[20:19] <Laurenceb> work then installing a bathroom
[20:20] <jcoxon> hehe cool
[20:20] <Laurenceb> its possible to do it when you're that tired.. if you have detailed planes and instructions carefully worked out beforehand...
[20:20] <Laurenceb> but not advisable
[20:20] <jcoxon> work = phd?
[20:20] <Laurenceb> yes
[20:20] <jcoxon> how much longer have you got?
[20:20] <Laurenceb> 3 years
[20:21] <Laurenceb> hows your degree going?
[20:21] <jcoxon> done :-p
[20:21] <Laurenceb> well done, what are you doing next?
[20:21] <jcoxon> working :-p
[20:22] <Laurenceb> ah lots of money :P
[20:22] <Laurenceb> but long hours
[20:23] <jcoxon> hehe going to get alot worse next week
[20:23] <jcoxon> we are massively understaffed
[20:23] <jcoxon> so i now have the job of 2
[20:23] <Laurenceb> :/
[20:23] <Randomskk> do you get the pay of two? :P
[20:23] <jcoxon> no
[20:23] <jcoxon> same hours :-p
[20:24] <jcoxon> well same paid hours
[20:24] <Randomskk> D: fun times
[20:24] <jcoxon> the fact i work longer is my fault for not being efficent apparently
[20:28] Action: Laurenceb tried watching TV
[20:28] <Laurenceb> not a good idea
[20:29] <Laurenceb> people being dropped into a lake roman gladiator style with horrendously corny production and cutting
[20:29] Action: Laurenceb facepalms
[20:32] <fsphil> not good is it
[20:34] <jcoxon> grrr i can't login to hysplit any longer
[20:37] <DanielRichman> Randomskk: is there a way to make monit wait a couple of cycles before trying to restart exim?
[20:38] <DanielRichman> ie is it possible to add the within x cycles clause to if failed host localhost port 25 protocol smtp then restart
[20:39] Laurenceb (~laurence@host86-152-35-235.range86-152.btcentralplus.com) left irc: Remote host closed the connection
[20:40] <Randomskk> I imagine so, but I'm not sure how
[20:40] <DanielRichman> I'm reading the manpage... there's a "noalert" command to disable certain emails but that's not really what we want
[20:40] <DanielRichman> also it'd be good to know why it's going crazy
[20:41] <Randomskk> that too
[20:41] LazyLeopard (~irc-clien@chocky.demon.co.uk) joined #highaltitude.
[20:41] <SpeedEvil> I watch very, very little TV these days.
[20:41] <SpeedEvil> Top Gear is about it. And Mongrels.
[20:41] <Randomskk> I watch one or two programs via iplayer or 4od
[20:41] <SpeedEvil> Oh
[20:41] <SpeedEvil> and Sky at Night
[20:42] <SpeedEvil> And have I got news for you too.
[20:42] <Randomskk> e.g. it crowd, the new Sherlock on BBC, dr who, some others occasionally
[20:42] <SpeedEvil> And a whole heap of radio stuff.
[20:42] <Randomskk> however dr who and it crowd have finished their season and sherlock only has one left
[20:42] <DanielRichman> Randomskk: also I mustn't forget to setup hardware raid monitoring. The website with the proprietary (eww) software to do so requires registration so I'm going to wait and see if ants already has an account (you have to register your card)
[20:42] <Randomskk> right, okay
[20:42] <Randomskk> worth setting up
[20:43] <DanielRichman> oh and http://github.com/danielrichman/pred
[20:43] <Randomskk> I didn't get too far with the python predictor, but far enough to think it could work and might be worth it
[20:44] <SpeedEvil> I recommend on Radio 4 - Material World, Home Planet, Thinking Allowed, Numbers In Action.
[20:44] <Randomskk> there is pyproj which can manage all the geometry - it can take a lat, lon, distance and bearing and give a new lat, lon, or it can take a lat, lon and give a cartesian position in meters
[20:44] <fsphil> watched a great movie last week on bbc two, "frequency asked questions about time travel" -- silly time travel comedy, good fun
[20:44] <Randomskk> the meters one is very easy to then manipulate for wind, you just add the wind vector on to each
[20:44] <DanielRichman> that would make things a bit easier
[20:44] <Randomskk> but the wind data is indexed by lat, lon so probably the most sensible is to stay with lat, lon and use the former tool
[20:45] <SpeedEvil> get_iplayer is great - stuff just turns up on my phone.
[20:45] <Randomskk> converting the cartesian wind vector to polar is no problem anyway
[20:45] <Randomskk> so that's easy enough
[20:45] <Randomskk> also the GFS tiles are 0.5 degrees big each side, so I think getting each tile as needed is probably practical
[20:45] <DanielRichman> well each tile isn't very big
[20:45] <Randomskk> it's not going to be very many
[20:45] <Randomskk> they're not small either - half a degree
[20:46] <DanielRichman> true; might be quite a few round trips
[20:46] <DanielRichman> is the nomads server quite good?
[20:46] <Randomskk> hard to say
[20:46] <Randomskk> I have a feeling the current setup involves making all those round trips
[20:46] <Randomskk> e.g. it requests each tile
[20:46] jcoxon (~jcoxon@host86-163-184-138.range86-163.btcentralplus.com) left irc: Quit: Leaving
[20:46] <Randomskk> separately
[20:46] <Randomskk> before predicting
[20:46] <DanielRichman> remember that for every balloon location you have to get 16 "cells"
[20:46] <Randomskk> I am not entirely sure though
[20:46] <Randomskk> oh yea, why?
[20:46] <DanielRichman> 4 axis
[20:46] <Randomskk> why do we interpolate it that much?
[20:47] <DanielRichman> lat long alt and time
[20:47] <Randomskk> yea, but the data is indexed like that, those four bits of data go into a single wind speed vector
[20:47] <DanielRichman> because for something that moves in metres, gfs is quite low resolution
[20:47] <Randomskk> interpolating doesn't increase our resolution
[20:48] <Randomskk> there is the HD dataset too
[20:48] <DanielRichman> but if we didn't then the balloon track would be a polygon (more than it already is)
[20:48] <DanielRichman> what are you suggesting, just use the wind vector from the nearest cell?
[20:48] <Randomskk> yea
[20:49] <DanielRichman> well it's not the >most accurate> thing as it stands, so ... I guess we'd have to try it
[20:49] <DanielRichman> PS: alt isn't actually an axis
[20:49] <DanielRichman> the ""altitude"" axis units is millibars
[20:49] <Randomskk> yea
[20:49] <Randomskk> it's a bit odd
[20:49] <DanielRichman> just to mess with you
[20:50] <DanielRichman> DODS seems to have very flexible requesting features. With teh right ranges, ew could just download data to cover the whole of the uk in one go
[20:50] <DanielRichman> any idea how big it would be?
[20:50] <Randomskk> not really sure. guessing a couple hundred meg based on current data usage
[20:51] <DanielRichman> couple hundred meg? so about 8 seconds.
[20:51] <Randomskk> depends how fast it comes in off their server
[20:51] <DanielRichman> oh yeah, true
[20:52] <DanielRichman> I was planning - if I decide to finish this C tidyup - to download gfs as needed and store each lat-lon 1x1 tile in its own file as binary data (structs) then mmap it in when I need it
[20:52] <Randomskk> sounds like a better idea
[20:52] <Randomskk> though to be honest you could probably afford to hold it all in ram
[20:53] <DanielRichman> well if it's just been downloaded then the kernel should still have it in ram; but that will also act as a cache
[20:53] <DanielRichman> then it's up to the kernel to put the 3gb to good use
[20:54] <DanielRichman> then; since it's in C; it would be relativly trivial to link a second binary that does batch downloads. Providing the flocking is done properly, that can run on a cronjob and speed up individual predicts
[20:54] <Randomskk> yes
[20:56] <Randomskk> man, doesn't matter how much cpu I throw at Bibble, it still wants moar
[20:56] <Randomskk> could do with 8 3ghz cores really
[20:56] <Randomskk> aww bugger I rename a folder and all its cached thumbnails are forgotton about
[21:02] <DanielRichman> bit of a pain that nomads doesn't support geogrid()
[21:08] Action: Randomskk experiments with summon-arm-toolchain
[21:09] <Randomskk> DanielRichman: you should ditch the eternally not stocked xmegas and migrate to stm32 =D
[21:09] <DanielRichman> Randomskk: I have some stm32 chips here
[21:09] <Randomskk> they're a load of fun tbh
[21:09] <DanielRichman> I'm going to start playing with them
[21:09] <DanielRichman> when I cba
[21:09] <Randomskk> yea fair enough
[21:09] <DanielRichman> however I think I've got some xmegas on their way
[21:10] <DanielRichman> so providing that goes OK then alien2 will go ahead as planned with an xmega
[21:10] <DanielRichman> allright let's see how fast nomads can spew out data
[21:10] Action: DanielRichman tries to download worldwide ugrdprs for the next week
[21:11] <DanielRichman> 412,119 282K/s in 1.4s
[21:11] <DanielRichman> oh that was an error page >.>
[21:12] <Randomskk> :P
[21:12] <DanielRichman> oookey let's try that again
[21:12] <DanielRichman> it's thinking about this one
[21:13] <DanielRichman> --.-K/s <-- I think I broke it
[21:13] <DanielRichman> date
[21:14] <DanielRichman> well... while it thinks about that one I'll try narrowing the parameters to just cover the UK
[21:14] <DanielRichman> Randomskk: RESULT!
[21:14] <DanielRichman> message = "subset operation failed; process exceeded time limit of 100 sec";
[21:14] <Randomskk> :D
[21:28] <DanielRichman> Randomskk: the limiting factor is very much the amount of ram & cpu on the nomads server
[21:28] <DanielRichman> I can only conclude that opendap isn't very efficient
[21:28] <DanielRichman> the data itself is only a few K
[21:29] <DanielRichman> but it all arrives in one go - god knows what the servers are doing? loading the whole response into ram, in advance?
[21:29] <Randomskk> really? a few k for what?
[21:29] <DanielRichman> hd[0:1:10][0:1:46][74:1:78][0:1:3]
[21:29] <Randomskk> dap is meant to be good but their servers could well be rubbish
[21:29] <DanielRichman> was ~50
[21:29] <DanielRichman> k
[21:29] <Randomskk> what's that in english?
[21:29] <DanielRichman> that's - as far as I know, 5 days, all 47 altitude levels, and a 2x2 latxlon square
[21:30] <DanielRichman> I could be mistaken
[21:30] <DanielRichman> when you consider that it's an array of float64s, that's quite a few floats
[21:30] <Randomskk> perhaps converting it to csv is where the waste comes in
[21:32] <DanielRichman> hmm my bad; that's not 10days; that's 1.25 days
[21:32] <Randomskk> what time resolution is it?
[21:32] <DanielRichman> 0.125 days
[21:33] <DanielRichman> if it's released every 6 hours we only need ~8 of the 64 items on the time axis before they're superseeded
[21:33] <Randomskk> well except that at any given time we want to be able to predict for the next week
[21:33] <DanielRichman> doh
[21:34] <Randomskk> I am compiling gcc with gcc, excellent
[21:34] <DanielRichman> You're not installing gentoo, are you? or a cross compile environment?
[21:34] <Randomskk> trying esden's summon-arm-toolchain
[21:34] <Randomskk> so it's gcc for arm-none-eabi
[21:34] <Randomskk> my old toolchain involved a lot of annoying hackery and codesourcery's gcc
[21:34] <Randomskk> if this works I'm much happier
[21:35] <DanielRichman> ok this blob: gfs_hd_12z.dods?ugrdprs[0:1:8][0:1:46][74:1:74][3:1:3] - 0.5x0.5 square, 1 day, all altitudes; takes 40seconds to download (39 spent waiting) and is 2,483bytes (nessie)
[21:35] <Randomskk> so basically their servers are the slow point
[21:35] <Randomskk> does the time you have to wait correlate linearly with the amount of data though?
[21:36] <Randomskk> or is it about a 40 second wait and then you get all your data
[21:36] <Randomskk> like, is it worse to make lots of requests
[21:36] <DanielRichman> that's the first one I've timed. before now I've only obsreved data or timeout. I'll see if it's linear
[21:37] <DanielRichman> hold on, that one was faster.
[21:37] <DanielRichman> I'll run a few and average >.>
[21:38] <DanielRichman> gfs_hd_12z.dods?ugrdprs[0:1:8][0:1:46][74:1:75][3:1:3] (1x0.5 square all alt 1 day) consistantly takes 32 seconds (I'll retry the first)
[21:38] <DanielRichman> PS: their data server is written in java
[21:39] <Randomskk> ah, no wonder then
[21:39] <DanielRichman> ok the 0.5x0.5 square has joined the other one in taking 32seconds
[21:41] <DanielRichman> a 4x0.5 square also takes 32 seconds ish;
[21:42] <DanielRichman> gfs_hd_12z.dods?ugrdprs[0:1:1][0:1:1][0:1:1][0:1:1] that is, asking for exactly one float, takes virtually no time at all
[21:42] GeekShadow (~Antoine@reactos/tester/GeekShadow) joined #highaltitude.
[21:43] <Randomskk> curious
[21:43] <Randomskk> well, summon-arm-toolchain has done exactly what you'd expect and created me an entire toolchain, but sadly not a build environment
[21:43] <DanielRichman> gfs_hd_12z.dods?ugrdprs[0:1:1][0:1:1][0:1:1][0:1:1] ~0.8s
[21:43] <DanielRichman> gfs_hd_12z.dods?ugrdprs[0:1:2][0:1:2][0:1:2][0:1:2] ~1.2s
[21:43] <Randomskk> it has even got libs and put them in proper kind of directories
[21:43] <Randomskk> hmm so
[21:44] <Randomskk> basically it kind of scales?
[21:44] <DanielRichman> gfs_hd_12z.dods?ugrdprs[0:1:4][0:1:4][0:1:4][0:1:4] ~ 2-3 s
[21:44] <DanielRichman> it would appear to
[21:44] <DanielRichman> also from what I observed of the [1][1][1][1], as soon as it connected, it returned
[21:44] <DanielRichman> ie. all of the 0.8 was latency
[21:45] <DanielRichman> so in theory there's a sweet spot.
[21:47] <DanielRichman> but if the blobs that we download get smaller and smaller, caching them efficiently becomes harder; there are a couple of things you could do.. sparse files, massive filesystem trees, proper database software...
[21:50] <DanielRichman> gfs_hd_12z.dods?ugrdprs[0:1:8][0:1:8][0:1:8][0:1:8] ~ 10s
[21:50] <Randomskk> and the most you can download?
[21:53] <DanielRichman> (...trial and error...)
[21:54] <DanielRichman> the closer you get to the goal each trial get significantly slower
[21:54] <DanielRichman> Also, it's hard to know if the time it takes is directly proportional to the number of cells requested
[21:54] <DanielRichman> gfs_hd_12z.dods?ugrdprs[0:1:16][0:1:16][0:1:16][0:1:16] 30s
[21:56] <DanielRichman> nomads don't have a fair usage policy or anything do they?
[21:56] <Randomskk> not that I'm aware of
[21:57] <DanielRichman> gfs_hd_12z.dods?ugrdprs[0:1:32][0:1:32][0:1:32][0:1:32] 1 minute 50 (about 10s downloading at 500K/s - I guess that's what nomads' b/w is)
[21:57] <DanielRichman> that is about the limit aslong as it wasn't lying when it said the limit was 100s
[22:07] <DanielRichman> linearly increasing the amount (in bytes) of data requested: http://pastebin.com/MSEjmLEg
[22:07] <DanielRichman> you can sort of trick yourself into seeing a correlation
[22:09] <Randomskk> heh
[22:09] <Randomskk> it kind of goes up
[22:11] <DanielRichman> indeed
[22:12] <DanielRichman> I think the point is that the prediction calculations arn't going to be the bottleneck
[22:12] <Randomskk> unless we have pre-downloaded this
[22:12] <DanielRichman> indeed.
[22:13] <DanielRichman> hopefully there's more than one box beind that nomads IP so we can parellelise grabbing the data and speed it up a bit
[22:13] <DanielRichman> but that means that a predictor that requests data when it needs it will run very slowly
[22:13] <Randomskk> yes
[22:13] <Randomskk> yes to the first, that is
[22:14] <Randomskk> the frontend is a high availability server that frontends for several servers
[22:14] <Randomskk> I believe you can connect to the other servers directly too if you want
[22:14] <DanielRichman> that would be useful, not much point contacting a server that is already toasting its cpu
[22:15] <jonsowman> hello :)
[22:15] <DanielRichman> hey there
[22:15] <DanielRichman> also worth noting that this is GFS_HD data. I haven't tried poking GFS
[22:15] <jonsowman> their idea of high availability is an interesting one
[22:15] <DanielRichman> :D
[22:16] <jonsowman> nomad1 and nomad3 both have the high def GFS dara
[22:16] <jonsowman> * data
[22:16] <DanielRichman> I wish they just provided a ftp server with a massive file
[22:16] <jonsowman> nomad5 sometimes does, but not reliably
[22:16] <DanielRichman> well... depends on how massive it is :P
[22:16] <Randomskk> jonsowman: yo
[22:16] <jonsowman> they all have the 1.0x1.0 degree data afaik
[22:16] <jonsowman> hello Randomskk
[22:16] <jonsowman> hows the weather station?
[22:16] <Randomskk> jonsowman: it is blinking its two LEDs =D
[22:16] <jonsowman> \o/
[22:16] <jonsowman> nice work :D
[22:16] <Randomskk> my desktop can't program it? for some reason?
[22:17] <Randomskk> will need to investigate
[22:17] <jonsowman> uh
[22:17] <jonsowman> what?
[22:17] <Randomskk> my desktop programs the other arm board fine
[22:17] <jonsowman> weird
[22:17] <Randomskk> but not the wx
[22:17] <Randomskk> but my laptop is okayt
[22:17] <Randomskk> same python script
[22:17] <jonsowman> i see
[22:17] <jonsowman> well at least that makes sense
[22:17] <Randomskk> only difference I can see is that the python script is using psyco since it's on my laptop
[22:17] <jonsowman> :\
[22:17] <Randomskk> but basically I have no idea
[22:17] <Randomskk> however
[22:17] <Randomskk> used esden's summon-arm-toolchain
[22:17] <Randomskk> it very nicely got and compiled me arm-none-eabi toolchain
[22:17] <Randomskk> and the stm32 lib and the libopenstm32 project
[22:17] <Randomskk> all excellent stuff
[22:18] <jonsowman> DanielRichman: in answer to your earlier question, no the datasets are not released at exactly the same time
[22:18] <jonsowman> Randomskk: :D
[22:18] <Randomskk> then I nicked the makefile from a libopenstm32 blink thing
[22:18] <Randomskk> the whole source for this blinker is 4K
[22:18] <DanielRichman> standard gfs seems to take much less time to get the same number of floats
[22:18] <DanielRichman> gfs_12z.dods?ugrdprs[0:1:16][0:1:16][0:1:16][0:1:16]
[22:18] <jonsowman> DanielRichman: they are usually within a 2hr window, but I've seen them miss sets entirely
[22:18] <Randomskk> it appears to be a lot more efficient at doing code than the old way
[22:18] <DanielRichman> 15 seconds
[22:18] <Randomskk> so this is all excellent
[22:18] <jonsowman> :)
[22:18] <Randomskk> also http://flickr.com/randomskk
[22:19] <jonsowman> ooo
[22:19] <jonsowman> that looks rather good really doesn't it
[22:19] <jonsowman> :)
[22:19] <Randomskk> until you zoom right in on the soldering on the ARM :P
[22:19] <jonsowman> that's not our fault
[22:19] <jonsowman> i blame farnell entirely for that
[22:19] <Randomskk> yes
[22:20] <Randomskk> ugh. why the heck can't my desktop program the thing
[22:20] <Randomskk> it's the same stm32loader.py script (exactly) and the same binary code to program (exactly)
[22:20] <jonsowman> that's really weird
[22:20] <jonsowman> back in 10, shower
[22:22] <Randomskk> ah, I can't get python-psyco on my desktop as it is 64bit
[22:23] <Randomskk> but psyco should make no appreciable difference
[22:23] <Randomskk> might be a voltage kind of thing
[22:23] <Randomskk> very hard to say
[22:23] <Randomskk> very weird.
[22:23] <Randomskk> notably, and most weirdly, the FTDI's TX LED is lit up dimly when connected
[22:23] <Randomskk> and pulses
[22:24] <Randomskk> this doesn't happen on the other stm32 board I have here
[22:25] <DanielRichman> Randomskk: I'm tempted to say that a dap library would be more effort that it's worth in C
[22:25] <DanielRichman> you just build a url with some numbers and it gives you an array of floats
[22:25] <DanielRichman> might be in network byte order, not sure
[22:26] <Randomskk> huh, so it's actually fairly easy.
[22:26] <Randomskk> fair enough
[22:26] <DanielRichman> liboc (C opendap) is massive; there's so much in it, I feel like I must be missing something
[22:27] <DanielRichman> brb
[22:28] Daviey (~Daviey@ubuntu/member/daviey) left irc: Excess Flood
[22:30] G8DSU (~chatzilla@cpc3-mort4-0-0-cust90.croy.cable.virginmedia.com) left irc: Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]
[22:32] Daviey (~Daviey@ubuntu/member/daviey) joined #highaltitude.
[22:33] SpeedEvil (1000@tor/regular/SpeedEvil) left irc: Ping timeout: 240 seconds
[22:34] <DanielRichman> worldwide GFS data (for the 3 vars we need) (in total) I reckon is 2.5GB, HD: ~18TB
[22:35] <Randomskk> a lot to get every 6hrs
[22:35] <Randomskk> and quite a bit pointless
[22:35] <DanielRichman> yep, I can forsee the winds over the south pole being v. useful
[22:35] <DanielRichman> especially when the predictor's maths breaks over ther
[22:35] <Randomskk> :P
[22:36] <DanielRichman> for the whole of the uk (square; so including a lot of sea) I reckon HD is only 50mb
[22:36] <DanielRichman> I could have slipped a calculator digit though
[22:36] <DanielRichman> just guessin'
[22:36] <DanielRichman> that's a more practical figure. And it would be even less for non-hd
[22:36] <Randomskk> yea
[22:37] <DanielRichman> those servers arn't giving it up without a fight
[22:37] SpeedEvil (~user@tor/regular/SpeedEvil) joined #highaltitude.
[22:38] <jonsowman> back
[22:38] <Randomskk> jonsowman:
[22:38] <Randomskk> uh
[22:38] <Randomskk> hi
[22:38] <jonsowman> sounds ominous
[22:38] <Randomskk> so, something is weird - the chip is working but the python is doing something quite odd
[22:38] <jonsowman> sup?
[22:38] <Randomskk> got the logic analyser out again
[22:38] <Randomskk> fantastic tool, that
[22:38] <jonsowman> :)
[22:38] <Randomskk> anyway I can see the python-triggered reset
[22:38] <jonsowman> what is it doing?
[22:39] <Randomskk> for some mreason at the same time as the reset it brings its TX line high for a short blip, then down low again
[22:39] <Randomskk> the reset is asserted for half a sec, then off for half a sec, then it sends some data thing
[22:39] <Randomskk> but the ARM replies before it sends that
[22:39] <Randomskk> but it reads 0, then later reads the right data
[22:39] <Randomskk> it's like there's a single byte in the buffer that's buggering it up
[22:39] <Randomskk> basically it's a bit odd.
[22:40] <jonsowman> weiiiird.
[22:40] <jonsowman> wonder what's going on there
[22:40] <DanielRichman> gfs -non hd- data for the whole of the UK is tiny
[22:40] <jonsowman> DanielRichman: how big?
[22:40] <jonsowman> estimate
[22:41] <DanielRichman> I wanna double check before I put a number on it
[22:41] <DanielRichman> buuut <1mb
[22:41] <jonsowman> really?!
[22:41] <jonsowman> surely not
[22:41] <DanielRichman> no...w ait... <10mb?
[22:41] <DanielRichman> I'll have to doublecheck the coords
[22:41] <Randomskk> well it's easy maths right - how many pressure levels, how much time, how many tiles
[22:41] <jonsowman> 10mb i could maybe believe
[22:42] jedahan (~jedahan@ool-45701ce6.dyn.optonline.net) joined #highaltitude.
[22:42] <jonsowman> from the evidence i've seen the HD data for preds is not particularly useful
[22:42] <jedahan> nice update to the predictor!
[22:42] <jonsowman> the non-HD one is just as good
[22:42] <jonsowman> jedahan: glad you like it :)
[22:42] <jedahan> just in time to - we are launching tomorrow morning
[22:43] <jonsowman> oo excellent
[22:43] <jonsowman> from where?
[22:43] <jedahan> and the github account makes it easy to clone and see whats going on
[22:43] <jedahan> florida, ny
[22:43] <jedahan> , usa
[22:43] <jonsowman> nice one :)
[22:43] <jonsowman> good luck!
[22:43] <jedahan> thanks
[22:43] <jedahan> we are participating in a hackerspaces in space contest and this is our test launch
[22:44] <jonsowman> ahh yes I've heard of it
[22:44] <DanielRichman> 65 times 26 alts 12 lats 20 lons 3 vars 8 bytes = 9734400 (9.2MB)
[22:44] <jonsowman> Randomskk: still programs fine on the laptop?
[22:44] <Randomskk> yea
[22:44] Daviey (~Daviey@ubuntu/member/daviey) left irc: Excess Flood
[22:44] <jonsowman> hum
[22:44] <jonsowman> DanielRichman: believable
[22:45] <DanielRichman> that's a massive rectangle around the UK
[22:45] <jedahan> gonna have full DVD quality video of the flight, and our total cost is $130. Some in our group just got their ham licenses and basic aprs equipment,so the next one might be much more interesting but not as cheap
[22:45] <jonsowman> is that a "rectangle" that encloses UK landmass?
[22:45] <jonsowman> oh right
[22:45] <jonsowman> hehe
[22:45] Daviey (~Daviey@bootie.daviey.com) joined #highaltitude.
[22:45] <jonsowman> jedahan: sounds excellent
[22:45] <DanielRichman> jedahan: stop taunting us like that, it's not fair :(
[22:46] <jedahan> DanielRichman, nothing stopping you from making your own parachute and payload container!
[22:46] <DanielRichman> airborne ham licenses
[22:46] <jedahan> ahh
[22:46] <jedahan> only took 30 days to go from no license, to getting the basic, general and extra licenses...
[22:46] <DanielRichman> nice :)
[22:46] <jedahan> the equipment is relatively inexpensive too - $35 for the microcontroller that does analog and digital encoding for aprs, then $80 for the power supply and antennae IIRC
[22:47] <jonsowman> lucky people
[22:47] <jedahan> is HAM only in the US?
[22:47] <jedahan> I mean, those licenses
[22:47] <DanielRichman> I'm UK intermediate; but even if I was UK full then I still wouldn't be allowed to fly my callsign. Ofcom don't let you use your license on airborne devices
[22:47] <Randomskk> jedahan: oh, we have licenses - I'm a UK Full, which is the highest we go up to
[22:47] <jedahan> ahh
[22:47] <Randomskk> but we are not allowed to operate airborne
[22:47] <Randomskk> so, we are limited to 10mW transmit power on 434MHz
[22:47] <jedahan> :(
[22:47] <Randomskk> your car's keyfob would drown us out
[22:47] <jonsowman> forced to use unlicensed bands
[22:48] <jonsowman> but otoh - 555km range with 10mW
[22:48] <jedahan> thats a shame
[22:48] <Randomskk> jonsowman: I think there might even be a new record now?
[22:48] <jonsowman> we've (in a UKHAS sense) not done badly at all
[22:48] <jonsowman> oh okay
[22:48] <Randomskk> but certainly 555km previously
[22:48] <DanielRichman> a new one?
[22:48] <Randomskk> on 10mW
[22:48] <Randomskk> crazy
[22:48] <jonsowman> didn't know that
[22:48] <DanielRichman> dang
[22:48] <Randomskk> DanielRichman: maybe not, can't remember, but thinking of timobel's flight which had a UK receiver
[22:48] <DanielRichman> mmm
[22:48] <jonsowman> oh that was fsphil trying to break it
[22:48] <jonsowman> and didn't hear a peep from the payload iirc
[22:48] <Randomskk> no, someone in england
[22:49] <jonsowman> oh okay
[22:49] <DanielRichman> wasn't there a rxer up north?
[22:49] <jonsowman> fair enough
[22:50] Daviey (~Daviey@bootie.daviey.com) left irc: Changing host
[22:50] Daviey (~Daviey@ubuntu/member/daviey) joined #highaltitude.
[22:52] <DanielRichman> what about this script?
[22:52] <DanielRichman> http://nomads.ncep.noaa.gov/cgi-bin/filter_gens.pl?dir=/gefs.20100731/06/pgrb2
[22:52] <DanielRichman> I'm not yet entirely sure what I just downloaded because it looks too small; and it's a "GRIB" file
[22:53] <DanielRichman> and there's also these http://nomads.ncep.noaa.gov/pub/data/nccf/com/gens/prod/gefs.20100807/12/pgrb2/
[22:53] <Randomskk> GRIBs are how we used to get the data
[22:54] <Randomskk> apparently GFS is better
[22:55] <jonsowman> i dont even know why GFS is better
[22:55] <jonsowman> it has the HD stuff, whereas I don't think GRIB does?
[22:56] <DanielRichman> http://nomads.ncep.noaa.gov/cgi-bin/filter_gfs_hd.pl
[22:56] <DanielRichman> ^^ grib hd
[22:57] Tkladol (~tkladol@094-134-045-062.dynamic.caiway.nl) joined #highaltitude.
[22:57] <jonsowman> oh
[22:57] <jonsowman> well ignore me then
[22:57] <jonsowman> :)
[22:57] <DanielRichman> well, I think so.
[22:57] <Randomskk> ugh, Logic still crashes all the bloody time on linux
[22:57] <Randomskk> annoying
[22:57] wonderchook (~wondercho@12.130.118.22) joined #highaltitude.
[22:57] <DanielRichman> there's a filter_gfs.pl script and a filger_gfs_hd.pl script so I would hope that they have GFS hd
[23:00] <Randomskk> the heck. looking at the traces, it seems like the laptop says "0x7F" once, where the desktop says "0x7F 0x00 0x01 0xFF"
[23:01] <Randomskk> but the ARM does the same in response
[23:01] <Randomskk> but my PC then doesn't say anything more, whereas the laptop kept talking
[23:01] <Randomskk> whaat.
[23:02] <Randomskk> and indeed the python code has the computer write 0x7F
[23:02] <Randomskk> so why does it appear to keep on doing stuff afterwards
[23:03] SpeedEvil1 (~user@tor/regular/SpeedEvil) joined #highaltitude.
[23:04] <fsphil> someone in north england did receive a few strings from timobel's balloon
[23:04] Tkladol (~tkladol@094-134-045-062.dynamic.caiway.nl) left irc: Quit: Java user signed off
[23:04] <Randomskk> at fairly significant distance iirc
[23:05] SpeedEvil (~user@tor/regular/SpeedEvil) left irc: Ping timeout: 265 seconds
[23:05] <SpeedEvil1> wasn't it ireland?
[23:05] <Randomskk> not in this case
[23:05] <fsphil> M0DTS
[23:07] <SpeedEvil1> ah
[23:07] <fsphil> I think he has a pretty mighty setup
[23:09] <Randomskk> I have heard stories of multiple 20+ele yagis on a rotator
[23:09] <jonsowman> :o
[23:09] <Randomskk> what on earth.
[23:10] <Randomskk> wish I had a scope already >.>
[23:10] <fsphil> I think he was the same who tracked the last few strings when BH4 went on its little tour of .nl
[23:10] <Randomskk> computer is being told write(chr(0x7F))
[23:10] <Randomskk> Logic sees 0x7F0001FF
[23:16] <Randomskk> jonsowman: lol what the hell
[23:16] <Randomskk> if I just
[23:16] <Randomskk> have it read a single byte
[23:16] <Randomskk> in the startup function
[23:16] <Randomskk> it all works
[23:19] <jonsowman> :S
[23:19] <jonsowman> i see
[23:19] <Randomskk> it seems to clear the 0x00 it otherwise reads when it wants 0x79
[23:19] <DanielRichman> do you have any idea why you switched from GRIB to GFS?
[23:19] <Randomskk> so that when it later reads, it gets 79
[23:20] <Randomskk> I believe RJW did the switch when he improved the predictor end of 09
[23:20] <Randomskk> I believe because it was better, or because pydap made it easier, or something
[23:20] <DanielRichman> maybe a thorough investigation is due
[23:21] <DanielRichman> either way; I'm off now; night
[23:21] <jonsowman> see you DanielRichman
[23:22] <Randomskk> seeya
[23:22] <Randomskk> jonsowman: anyway yay, blinkey LEDs programmed from desktop
[23:22] <Randomskk> excellent
[23:22] <jonsowman> yay
[23:22] <Randomskk> this libopenstm32 seems fairly decent
[23:23] <Randomskk> apparently not yet feature complete but they are working on it
[23:23] <Randomskk> but it's goo
[23:23] <Randomskk> good*
[23:23] <Randomskk> it is much, much, much better than the ST libraries
[23:23] <Randomskk> it's like
[23:23] <Randomskk> gpio_set_mode(GPIOC, GPIO_MODE_OUTPUT_2MHZ, GPIO_CNF_OUTPUT_PUSHPULL, GPIO6);
[23:23] <Randomskk> gpio_set(GPIOC, GPIO6);
[23:23] <Randomskk> gpio_clear(GPIOC, GPIO6);
[23:24] <Randomskk> c.f. http://github.com/randomskk/STM32_SkeletonProject/blob/master/main.c
[23:24] <Randomskk> GPIO_InitStructure.GPIO_Pin crap, etc
[23:24] <Randomskk> with libopenstm32: http://github.com/esden/libopencm3/blob/master/examples/stm32-h103/miniblink/miniblink.c
[23:27] <jonsowman> :)
[23:39] RocketBoy (~steverand@217.47.75.8) joined #highaltitude.
[23:42] <Randomskk> jonsowman: http://github.com/randomskk/SelWX/tree/master/embedded_code/
[23:42] <Randomskk> jonsowman: btw we should totally shove a mini stm32 on the night launch
[23:42] <Randomskk> one of the 48 pin babies. they are great fun, when they solder properly
[23:50] wonderchook (~wondercho@12.130.118.22) left irc: Remote host closed the connection
[00:00] --- Sun Aug 8 2010