2008-04-10.log

--- Day changed Thu Apr 10 2008
-!- kanzure [n=bryan@cpe-70-113-54-112.austin.res.rr.com] has quit ["Leaving."]01:33
-!- kanzure [n=bryan@cpe-70-113-54-112.austin.res.rr.com] has joined #hplusroadmap07:18
-!- Enki-2 [n=weechat@c-71-234-190-248.hsd1.ct.comcast.net] has quit [Read error: 104 (Connection reset by peer)]08:46
kanzurehttp://heybryan.org/mediawiki/index.php/2008-04-10 for some news/notes/stuff08:50
kanzureWill somebody please check the http://perlmonk.org/ chatterbox in 20 minutes to see if anybody answered my latest message? I need to run.08:50
-!- kanzure [n=bryan@cpe-70-113-54-112.austin.res.rr.com] has quit ["Leaving."]08:50
-!- Enki-2 [n=weechat@c-71-234-190-248.hsd1.ct.comcast.net] has joined #hplusroadmap08:51
-!- Aulere [n=nevermor@131.229.176.252] has joined #hplusroadmap17:35
-!- kanzure [n=bryan@cpe-70-113-54-112.austin.res.rr.com] has joined #hplusroadmap17:39
kanzureHi Aulere.17:41
Aulerehi kanzure17:52
-!- Aulere [n=nevermor@131.229.176.252] has left #hplusroadmap []17:57
-!- Aulere [n=nevermor@131.229.176.252] has joined #hplusroadmap17:58
-!- faceface [n=dmb@p57A0C64A.dip.t-dialin.net] has joined #hplusroadmap18:04
facefacelets breed baby!18:04
facefacebtw, how are you getting on with your BiO.CC account?18:09
kanzureDan, hey there18:19
kanzureSorry for my lag18:19
kanzureI've been logging in and checking out some of the folders. Lots of interesting stuff in there.18:19
kanzureI don't understand, though. What's going on there? How many people, who are they, why? etc.18:19
facefacehehe - its confusing I know18:21
kanzureand I checked out the stats, KIGG (the parent govt body) has $100 million in funding18:21
kanzureBut it doesn't look like there's much money being used by KOBIC18:21
kanzurewhat's going on?18:21
facefacebasically ... well... perhaps there is no basically.... Jong went to korea to be professor... then he quit and went to another place ... then he quit again I think... now he is the director of KOBIC 18:21
facefacedon't know.18:22
kanzureI'm glad I know the director of KOBIC, but I don't know what that *means* - how much influence and push does he have?18:22
facefaceJong is good at leaving a trail of chaos behind him... bio.cc is the center of his chaotic world ;-)18:22
kanzureCould he get some of my bio projects funded? (there's not much in way of costs, of course, just need a lab)18:22
facefacekanzure, not sure... don't know what KOBIC does really... the papers carried a story in korea when he quit his post at KAIST - it was unheard of for a professor to quit and go to 'industry' so KOBIC is kinda like Lion biosciences vs EBI type thing18:23
kanzurehm18:23
kanzureI mean, from his websites and biosophy,18:24
kanzureit really seems like he digs my 'DIY biohacking' ideas18:24
facefacekanzure, you can surely ask about funding / collaboration. He payed for me to visit him in korea some time ago... If I can I would like to go again this summer.18:24
kanzureI see.18:24
kanzureI think he would really enjoy helping me out on the biohacking kit18:24
facefaceyeah... biosapien... 18:24
kanzurewhat I need to do is publish videos and documents that describe the *exact* steps to make the machinery18:24
facefacekanzure, how long would you need / how long a summer holiday do you have / how much would it cost... I am sure Jong would be interested, but he often complains about 'the system' where he is, so his 'push' is a constant struggle18:25
facefaceit seems... I don't really know.18:26
kanzurehm18:26
kanzureMy summer is June 1st to August 1st (basically)18:26
kanzureand while I know that I would not be able to get everything done within that time, I can focus on a single project - such as DNA sequencing.18:27
kanzure(I am thinking of the STM version of DNA sequencing.)18:27
kanzureit requires some thought obviously18:28
kanzurebut I think that the idea of releasing specific videos on doing these things would be a good push18:28
kanzure*show them* how to use the bioinformatics databases (you know, the screen recorders)18:28
kanzurethen show them simulations and so on (obviously you cut out the hour-long pauses as you figure out what to do ;))18:28
kanzureand then direct, elegant implementations into wetware18:28
facefacekanzure, jong has always talked about wanting to build a mechanical sequencer (as we discussed) - also he wants to found a company to exploit such a product. I am sure he would be delighted if you proposed such a project... you would need to be very focused though with lots of planning.18:31
kanzurehaha18:32
kanzurewell, you know me18:32
kanzure<-- intense18:32
facefacecool18:33
kanzureremember my site?18:33
kanzurehttp://heybryan.org/18:33
facefaceindeed18:34
facefacetake care. Good luck with your projects. good night.18:39
kanzureg'night18:40
-!- faceface [n=dmb@p57A0C64A.dip.t-dialin.net] has quit ["So long, and thanks... probably."]18:42
fennintense but not focused19:14
kanzurefenn: http://www.w3.org/TR/wsdl19:15
kanzurebah, I'd say I am focused19:15
fennfocused on a broad subject, maybe19:15
kanzureah, well, I suppose you may mean that in the sense that I don't just isolate myself completely to get a single item done at a time19:16
kanzurebut this is mainly because I am working off of a strategy to make sure I don't pick something that doesn't work19:16
fennum, wsdl, relevant? how?19:16
kanzurethe perlmonks told me that wsdl might be my way of doing formalized usage requirements19:17
fennO-o19:17
kanzurebut I don't get it19:20
kanzurelet's just do slip19:21
kanzureor xml19:21
kanzuredo we want to rip apt-get directly, or write our own client from scratch?19:21
kanzurehttp://en.wikipedia.org/wiki/Advanced_Packaging_Tool19:23
kanzurelooks like we can do autogenix with just bash scripts19:25
fenni'd rather use python19:27
fennbash is really hard for anything non-trivial19:27
kanzureI agree19:27
kanzureI'd rather do perl, frankly, because of the larger historical context of packages and so on19:28
fennoh pff19:28
kanzurewait19:28
kanzureit doesn't matter that much19:28
kanzurewe can do bindings back and forth19:28
fennwe can use apt for software packages19:28
fennright, bindings are a great idea, lets do more of those :)19:28
kanzuresort of - we'd need to hack apt for those meta-files19:28
fennhmm? apt would be a sub-set of autogenix19:29
kanzureyes19:29
fenna software package would appear to apt as a leaf node, but in autogenix it would just be part of a tree19:29
fennor is it a root? i always get confused19:29
* fenn would like some decent package visualization software19:30
fenndependency^19:30
kanzurethere are a few distinctions to make for software19:30
kanzuresome software would do simulations, others would be parsers and generators, converters, etc.,19:31
kanzurein the sense of dependency-loops in the self-replicator design, I doubt software would matter that much; once there's an ability to do software storage, all software can be implemented on it19:31
fennsure, simulator/test/parse would be specified somehow19:32
kanzureeh19:32
kanzurethis sounds like Cantor's mathematics19:33
kanzurethat the number of identifiers is the same as the number of possible programs19:33
kanzureor something like that19:33
* fenn googles19:33
kanzureit's ignorable at the moment19:33
kanzureI haven't thought about whether or not we would specify the structural relationships between the software packages19:33
kanzureI hear that this is a big problem in apt, they only have a few ways to specify dependency19:33
kanzureand dependency-resolution in their system is an even bigger problem19:34
kanzure*dependency-error resolution19:34
fennlol n*infinity19:34
fennmy math teacher wouldn't have any of it19:34
kanzureokay, so I guess there's only a limited number of relationships that we want to express anyway19:36
kanzurethis is just dependency or whatever - "download this at the same time in order to make it run" dependency stuff19:36
kanzureother stuff like "is an alternative to" doesn't have to be specified on the package level ... rather, a larger package would reference all of the packages that are of the same type, and a person would make that package file if they were so inclined19:36
kanzure(plus reverse links of course)19:36
kanzures/reverse links/backlinks/19:37
fennyou can do aliases for dependencies19:37
fennin apt19:37
fennwell, dpkg really19:38
kanzuresounds like ##REDIRECT to me19:38
fennright19:38
fennwhy is redirect discouraged?19:38
kanzurewhich reminds me -- it'd be really nice if we can have CVS integrated into the backend for the wiki ... CVS for mysql or something, and then just make sure that it's editing the wiki tables. That'd be really helpful.19:38
kanzureis it?19:38
fennyeah you're supposed to send 405 or something19:39
kanzureoh19:39
kanzureit's because redirect is in the HTTP header19:39
kanzurewhereas javascript/html redirect is more boggy stuff19:39
kanzureI'm all for 405 :)19:39
kanzureapache mod_rewrite etc19:39
fennsure19:39
kanzureand mod_redirect IIRC19:39
fennmod_rewrite is bitchun19:39
fenn(when it works)19:40
fennCVS is old, but i get your drift19:40
fenni like git because it can track changes between files19:40
kanzureI refuse to let this be an HTTP-only interface ordeal, where autogenix would have to go through HTTP and HTML to get stuff ... how absurd.19:40
fennlike, if you move code from package a to package b19:40
kanzureI have no preferences for versioning systems, so I'm good for your preferences19:41
fennwhat's wrong with http? it gets through firewalls and downloads packages19:41
kanzurewait, http is ok19:41
fennyou mean UI?19:41
kanzurebut not html to represent the wiki page only19:41
fennright19:41
kanzurehm19:41
kanzureI wonder how we would opt to do that19:41
kanzurewould we want to dump the package as a wiki-compatable-format19:41
kanzureor would we want the wiki to retrieve the file from another table in the db19:42
fennwiki is a nebulous concept :)19:42
kanzureand then do its own markup19:42
kanzuretrue19:42
kanzureI was hoping for mediawiki just to do the status quo thing19:42
fennit doesnt have to be a markup language19:42
fennfuck mediawiki19:42
kanzurexml is markup19:42
kanzurealthough slip isn't19:42
fenner, uh, i mean a display markup19:42
kanzureah19:42
kanzuregoing over it:19:43
fennok, the 'wiki' software's job is to take our XML markup and transform it into some pretty HTML19:44
kanzureohhh19:44
fenni'm sure there's hundreds of things out there that can do that19:44
kanzurewhat if our XML is compatable with our standards + wiki at the same time19:44
kanzureor at least we can have a wiki extension to interpret it19:44
fennyeah, i thought you had come to that conclusion already :)19:44
kanzurefor example, in mediawiki you can do <youtube> tags, so same --19:44
kanzuresure, I'll pretend like I did19:44
fennso.. there's more to it than just xml19:45
fennalso cad files, binaries, assorted nastiness19:45
kanzurexml must reference those files, yes19:45
kanzurei.e., "<dataFile>/path/to/my.cad</dataFile>"19:46
fenni'm worried about version control systems and large quantities of binary files19:46
fennit could turn into a mess quickly19:46
fennbut i dont have any suggestions19:46
kanzureare you worried about messy dir structures, or too much used space?19:46
fennspace, inability to track changes19:46
fennmore like, bandwidth for mirroring19:47
kanzurere: space, periodic package pruning? with also backup and mirroring to offsite storage19:47
kanzurehm19:47
fennwe could just say, 'dont do that'19:47
fenni.e. all illustrations in svg19:48
fenncad files in ascii19:48
fennopen source cad doesn't exist yet.. wtf are we going to do about that19:48
kanzureare you sure?19:49
kanzurethat sounds unlikely19:49
fenn3d stuff19:49
kanzurehttp://blender.org/19:49
fennnot cad19:49
kanzureI've never really used CAD19:49
fennuh, it's hard to tell the difference until you actually try to build something with it19:49
fennbasically, blender is like smooshing modeling clay around19:49
kanzureright19:49
fennand cad is like, programming a computer19:50
fennthere are various degrees in between too, like rhino19:50
kanzurenot working with crappy blender-interfaces?19:50
fennnothing to do with the interface19:50
kanzureI think blender allows python scripts now though19:50
kanzureso that you can program it to make models 19:50
fennthere have been attempts to create a cad interface for blender, miserable failure19:50
kanzurehere's what I am thinking - 19:51
kanzure(1) write up some autogenix code? do we want to steal from apt directly? what do we want it to be able to do? this is dependent on the package definition format; so really we just need a way to bootstrap the downloading of a new version of a script to work with the db19:51
kanzure this is dependent on the package definition format; so really we just need a way to bootstrap the downloading of a new version of a script to work with the db19:51
fennbasically, a 3d modeler doesn't store high level information, or real-world technological information19:51
kanzure(2) write up the meta-information / file format io specs or DTD (investigate?) ... this is the usage-statement stuff that we were talking about19:52
kanzure(3) wiki/versioning combo investigations19:52
fennin a first draft, i want autogenix to verify: units, connectors, signal types, software compatibility19:53
kanzurehow does apt do it? locally?19:53
kanzureI think apt downloads the overhead database and then does local scanning19:53
kanzureand then only calls the server to get the package19:53
fennapt downloads the control files, right19:53
fennactually it downloads diffs which are updated regularly (somehow)19:53
kanzurehuh19:54
kanzurewell, that would work well with #319:54
fennwe could do that with git19:54
kanzuresince wikis and versioning systems both wrok with diffs19:54
fenngit diff <db root>19:54
kanzureneat19:54
kanzureokay, so you want autogenix to be able to interpret package files19:55
kanzureI wonder if it's worth throwing that into another program19:55
kanzureapt-get = autogenix, apt-check for verification/validation, 19:55
fenni know you're trying to prod me into writing code, but i gotta do my taxes tonight :\19:55
kanzureI'm trying to prod you into specifications19:55
kanzureI can do code, if I know what I want19:55
kanzurehehe19:56
kanzurethere's also the 'usage message' formalization lib - this will come later when we start throwing in some software into the database, I guess, don't think it's that relevant to these bootstrap programs really19:57
fennapt-check just verifies that the db is consistent19:57
fenn'lintian' is one program that validates a debian package19:58
kanzuredoes apt-check actually exist ?19:58
fennno, i was thinking apt-get check19:58
kanzureah19:58
kanzuredidn't know19:58
fennusage message? for software or hardware or package management?19:59
kanzurethe usage-message stuff is for, say, 20:00
fennor just documentation in general?20:00
kanzurepackage re: chairs, references some software that simulates chairs20:00
kanzurechairSimulator requires certain input data20:00
kanzurelike a 3d object file20:00
fennan API?20:00
kanzureyeah, 20:00
kanzureremember the inline code we were discussing?20:00
fennthat's usually too much to pack into a man page20:00
kanzureright20:00
kanzurea man page is plain text anyway20:00
kanzureI'm thinking that this would be like "--help2" -- any program could call this, and get back formal information that says parameter 'testParam' requires file format of type ID 3940141 and then skdb has some information on that format20:01
kanzure*any person could call this20:01
kanzuredon't know if programs will, but if they do, they might find a way to use it20:01
kanzuremore importantly this xml spec would be available next to the program too20:02
kanzureso that it can be read without executing the program20:02
kanzurewe could do away with that idea entirely if we have to20:10
kanzureI'm pretty sure it's useful20:10
kanzureas it is, with apt-get, you could get a physics simulator, but not be able to interpret the results because nobody tells you what the file formats really are20:11
fennhttp://en.wikipedia.org/wiki/Table_of_mathematical_symbols20:16
fennlots of programs dont generate output files20:17
fennor if they do, its intended as a debug or strictly for hackery20:18
fenni do like the unix philosophy though20:18
fennit's almost mandatory for something this large and complex anyway20:18
fennautomated man-page reading would be powerful, but i dont even know where to begin (and it may be beyond the scope of what we're doing)20:20
fennit's not that hard to tell what kind of file you're reading, as usually there's a limited number of choices anyway20:22
fenncase "slip-xml": parse_slip(); case "stl" parse_stl(); etc20:23
fennsay you had a connector modeled in stl20:23
fennlike a wall plug or something20:23
fennthat stl file would be given a name, and could be referenced as a member of the electricity module20:24
fennand linked to the autogenix 'unit' describing that interface?20:25
fennmaybe interface needs to be a separate parameter for describing functionality20:25
fenn(instead of being glommed onto units)20:25
fenn110V 60Hz 15A are the units, but the interface is the familiar 0.o plug20:27
fennsoftware is another ball game though20:27
fennhmm. have to get wave-form in there too i guess20:28
fennsinusoidal is not the same as what most inverters put out, and then there's 'magic sinewaves' and so on20:29
fennalso, all these values want tolerances20:30
fennwe could re-name it 'technological interface database with diagrams included' :)20:30
kanzureautomated man page reading *will* be limited. You can only write in terms of text what the parameter is *for*, but you can specify what to give it :)20:34
kanzurefor the connector in that case - those units would have to be variablized based off of the material specifications, right?20:36
kanzurethink of it like a typical electronics lab20:36
kanzuremy high school electronics shop has this huge wall of compartments20:36
kanzureof tens of thousands of components20:36
kanzurein terrible organization, of course20:36
-!- fenn_ [n=pz@adsl-75-62-113-75.dsl.bltnin.sbcglobal.net] has joined #hplusroadmap20:37
-!- Topic for #hplusroadmap: http://heybryan.org/ http://heybryan.org/mediawiki/ http://heybryan.org/mediawiki/index.php/Self-replication | krebs is now servicing the channel. try !help20:37
-!- Topic set by krebs [] [Mon Apr 7 03:08:21 2008]20:37
[Users #hplusroadmap]20:37
[ Aulere] [ Enki-2 ] [ fenn ] [ kanzure] 20:37
[ drazak] [ epitron] [ fenn_] [ krebs ] 20:37
-!- Irssi: #hplusroadmap: Total of 8 nicks [0 ops, 0 halfops, 0 voices, 8 normal]20:37
-!- [freenode-info] if you need to send private messages, please register: http://freenode.net/faq.shtml#privmsg20:37
-!- Channel #hplusroadmap created Sat Mar 22 15:44:12 200820:37
kanzurespecifying the units like that seems kinda weird though20:37
kanzureat first I guess we can have tens of thousands of packages specifying different resistors20:38
-!- Irssi: Join to #hplusroadmap was synced in 36 secs20:38
kanzure(of course, we wouldn't do that by hand - we'd eat some digikey.com data sheets or something)20:38
kanzure(or even have an automatic cron job to update based off of digikey.com database queries, as long as we keep up with their HTML format output specs)20:38
kanzurebut on the other hand, if we come upon the software to model the materials used to make those resistors20:38
kanzurethen that's really, really useful20:38
kanzureso that we can then just have one package that can provide a conventional resistor20:38
kanzureand then we can pass some inputs to it to get the output for a type of resistor that we want20:39
kanzureeither way should wrok20:39
kanzure*work20:39
fenn_i got a decent idea; units would be named arguments to the interface class20:40
kanzuredefine interface class20:40
fenn_a code object that references all the cad files and whatnot20:41
-!- fenn [n=pz@adsl-76-248-68-178.dsl.bltnin.sbcglobal.net] has quit [Nick collision from services.]20:42
-!- You're now known as fenn20:42
kanzureoh, interesting20:42
fenndigikey allows parametric search, so you just search for what's specified and get a list of components that match20:42
kanzurecool20:43
kanzureokay, I wasn't expecting your interface class idea20:43
kanzurewhat I was thinking of was that the package-format could have a way for the programmer to list related files or whatever20:43
kanzureso that you would drop a cad file into a project if it's relevant20:43
kanzure(just link to it, actually, instead of multiple copies)20:43
fennhmm, i was thinking the class would usually be specified in a standard interface library, like NEMA or some crap, but yeah you could make your own interfaces20:44
fenni'd like to avoid the scenario where someone thinks they have all the software needed for building a package, but then at (software run) (tech build) time it turns out there's a dangling reference to some file20:46
kanzurethat's what the validator and ref-checkers would be for20:46
kanzureor20:46
kanzurewe have people devote themselves to making sure a package they throw up really, truly works20:46
kanzureand we can say "this *will* work if you do a full download"20:46
fenni think i'd rather trust the machine :)20:47
kanzuresure20:47
fennand devote the energy toward making sure the validator works20:47
kanzurewell, how about this20:47
kanzureif the validator finds that there's something missing in the packages, and has to go search the database20:47
kanzurethe person using the software doesn't necessarily know what interface is needed to make it work20:47
kanzurethey will get a list20:47
kanzureand perhaps there will be only one program that they can choose from, if so that's great20:47
kanzurebut otherwise they might be clueless20:47
kanzurethey could either do trial and error20:47
kanzureor they could generate a debug report or whatever20:48
kanzureand then go ask somebody who might have a clue20:48
kanzureand if a package *is* chosen, then this should be an auto-generated use case20:48
kanzureand uploaded to the database for analysis/use/whatever20:48
fennhurr.. the interface should be specified enough so that it truly doesnt matter which one they pick20:48
kanzurewell, think about the tools that do different things to files20:48
kanzurefor example, html2pdf versus frontpage20:49
kanzureboth have html input specs20:49
* fenn was thinking text editors20:49
fennuh, lets just leave frontpage out of this okay/20:49
kanzurehaha =)20:49
kanzurejust trying to think on my feet20:49
fennyou cant say 'look, broken software exists and people use it, so it is a possibility'20:50
kanzureso the solution is not at run-time or download-time20:50
kanzurebut rather approval-time20:50
kanzurewhen we are adding the package into the database20:50
fenncould be both20:50
kanzurewith run-time and/or download-time you get into territory where you get ambiguous software suggestions though20:51
fennthe database could get out of whack, then the user unwittingly downloads broken packages and believes they are ok20:51
kanzureyep20:51
fennwhy ambiguous?20:51
kanzurelike I said: html2pdf v. the unmentionable program20:51
kanzureboth take HTML input20:51
fennhtml isnt a specification20:51
kanzurestl v. blender20:52
fennuh, blender isnt a specification :)20:52
fennits a core dump20:52
kanzuredot blender files20:52
kanzureoh?20:52
kanzurehah20:52
fennsrsly20:52
kanzurewhat the hell are they smoking?20:52
kanzureI guess it could be sorta easy20:52
fenni think you cant even load different versions of blender files into blender20:52
fennok, so, stl vs. collada20:53
fennboth well defined, and incompatible20:53
fennso no reason the package should pick the wrong one20:53
kanzurethose are file formats20:53
fennstl 1.0 vs stl 2.020:53
kanzurenono20:53
kanzurelet's suppose we have20:54
kanzurestl-program-120:54
kanzurestl-program-220:54
kanzurethus, ambiguity 20:54
fennok, stl-program-1 outputs stl version 1.0 file format?20:54
kanzurethe cur package does not care what the output of whatever-stl-prog is20:54
kanzure(let's just say)(20:55
fennit should care what the output is, that's its job20:55
kanzureokay, good point20:55
fennif it's underspecified, that should be fixed20:56
fennwould like a piece of software to detect that sort of programmer error20:56
fenni.e. if a program can read stl2.0 then it can read stl1.0, doesnt mean the other way around will work20:57
fennsounds boring to code though20:57
fennthere really should be some big database describing all the file formats, everywhere20:58
fennor a language that the people who create file formats can decribe their format with20:59
fenn<.rantD[D[D[D[D/>20:59
* fenn kicks his wifi21:00
* fenn grumbles about EXPRESS21:03
-!- Aulere [n=nevermor@131.229.176.252] has quit [Connection timed out]21:09
kanzureyep, format db indeed21:31
kanzurewe will just keep the format db separate from skdb (even though it is a part of it, yes)21:32
kanzureso that it too can be mirrored and backed up21:32
drazakkanzure: any luck with those books?21:57
-!- Aulere [n=nevermor@131.229.176.252] has joined #hplusroadmap22:15
kanzurehttp://en.wikipedia.org/wiki/Rick_Tumlinson <-- This guy funded the Mars Society and did FINDS. He funds space-tech groups. But the trick is that you don't find him, he finds you, or else he rejects your application. 23:19
-!- Aulere [n=nevermor@131.229.176.252] has quit ["time for bed"]23:50

Generated by irclog2html.py 2.15.0.dev0 by Marius Gedminas - find it at mg.pov.lt!