Camelia, the Perl 6 bug

IRC log for #mojo, 2011-12-03

| Channels | #mojo index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:00 vervain Or... again if there's something that I can go read... just let me know that too <please>
00:00 japhb vervain, FWIW I just a couple minutes ago read about a default 5 MB limit on HTTP request parsing, for security reasons
00:00 japhb lemme see if I can find the right tab
00:00 japhb http://mojolicio.us/perldoc/Mojoliciou​s/Guides/FAQ#What_does_the_error_22Max​imum_message_size_exceeded2E22_mean3F
00:03 japhb Does Mojo::UserAgent work with (or have an equivalent replacement) for Mechanize and Test::Mechanize?
00:03 japhb Oh wait, it looks like Test::Mojo is that
00:03 sri mechanize is old school, we do css3 selectors
00:04 sri you need to try Mojo::DOM
00:04 japhb sri, you have no love for the old school, do you?  ;-)
00:04 sri just nostalgia
00:05 sri (a little)
00:05 japhb *chuckle*
00:08 japhb OK, between Mojo and Mojolicious, there's a LOT of stuff here.  Given the FAQ you linked above, that you have replaced the basic HTTP layer ... how far down did you go?  Where did you stop in the reimplementation?  Is everything built directly on IO::Socket etc.?
00:09 sri IO::Socket::INET, IO::Socket::SSL, IO::Socket::IP, yes
00:09 japhb Hmmm.
00:09 sri it's a custom event loop layer with pluggable backends
00:10 japhb And Mojo does all the HTML and XML parsing/creation too?
00:10 sri we support EV out of the box for example
00:10 * japhb nods thoughtfully
00:11 sri would be scary if a framework didn't do HTML creation ;p
00:11 vervain sri: am I in the weeds with this error then?
00:12 sri vervain: i'll take a look at the chrome error, no promises that i can fix it though, needs to be replicated in a unit test first
00:13 vervain That's my worry though... if curl always seems to 'get it' but chrome/ff don't I don't know what to do.
00:13 vervain What I mean is that if curl works then Test::Mojo might too.
00:13 japhb sri, I meant, did you create your own templating and/or programmatic HTML generation module(s) as well?  Or do you use TT or suchlike?
00:13 vervain I guess I'll try it and see.
00:14 sri japhb: we have a ERB clone and support TT through plugins
00:15 japhb Ah, no wonder I didn't recognize ERB immediately, it's Ruby.  :-)
00:15 sri main reasons for implementing our own ERB clone were good error messages and new syntax experiments, like block helpers that get compiled to perl closures
00:15 marcus Sri: mojo should have a http request recorder. For teste
00:15 marcus :(
00:16 sri marcus: totally
00:16 marcus :) damn hard to thumbtype on the ipad.
00:16 sri marcus: luckily all events to make that happen are already there
00:17 marcus Sri: how thoughtful =)
00:17 sri i've actually just used the write event myself to debug the chrome issue :D
00:17 sri it's so damn useful
00:19 sri Mojo::IOLoop->stream($self-​>tx->connection)->on(write => sub { my ($stream, $chunk) = @_; say "WRITE: $chunk"; });
00:19 japhb Is Mojo currently used by any other projects than Mojolicious?
00:19 sri be more specific
00:19 marcus I use it in some stuff
00:20 marcus Like the put.io module
00:21 sri Mojolicious is really much like the old LWP, a toolbox you pick the parts you need from
00:21 japhb Hmmm.  I suppose I mean two things: 1. Are there any other frameworks that used Mojo as a basis, but replaced the Mojolicious layer(s)?  2. Are there any NON-framework projects that are using Mojo as their HTTP engine.  Like e.g. a distributed testing system, say.
00:21 marcus I know my client uses the ua for their ws client.
00:21 sri 1. No 2. Yes
00:22 sri there have been frameworks on a slightly higher layer though
00:22 japhb marcus, good data point, thank you.
00:22 japhb sri, oh?
00:22 sri Mojolicious::Lite alternatives built around Mojolicious the same way
00:22 japhb Oh, I see.
00:22 sri it's more fun that way
00:25 japhb Does Mojo/Mojolicious have a preferred paradigm for using multiple cores on a single server, assuming Linux as the OS?
00:25 sri see hypnotoad
00:26 japhb BTW, thank you for being patient with all my questions.  Your answers have been very helpful.
00:26 sri sure thing, i actually like discussing design :)
00:27 japhb DDG is giving me an odd result for hypnotoad.  Somehow I don't think you're talking about Tokomaks.
00:28 vervain Sorry if this is in the docs... but how do I run the test suit in the mojo repo?
00:28 sri http://mojolicio.us/perldoc/Mojol​icious/Guides/Cookbook#Hypnotoad
00:28 marcus http://blog.moam.be/
00:28 marcus Vervain: make test
00:28 purl make test is chugging along fine in OpenBSD.
00:29 vervain Ah ya doh... sorry... I've been at this for 13 hours... and I've started pouring beer over the coffee. :-)
00:29 vervain I'll be in bed soon... but it's hard to walk away from a problem. ;-)
00:29 sri also http://mojolicio.us/perldoc/Mojoli​cious/Guides/Cookbook#Faster_tests
00:30 vervain How about running a one off test?
00:31 marcus Prove
00:38 vervain G'Night folks... the little people want me to decorate a tree for some reason. ;-)
00:39 vervain Thanks a bunch for your assistance... much appreciated!
00:46 tempire TO THE MOJOLICIOUS!
00:47 sri vervain: problem seems to be that we are closing the connection before chrome/ff read the response
00:48 sri not sure this can be handled better
00:49 sri oh i know
00:50 sri we send the response and it keeps writing the request body, we close the connection and it gets a broken pipe
00:51 sri ignoring the response and reacting just to the broken pipe
00:51 vervain BTW - on a quick whim I just tested IE... same as the other two.  5MB works as expected 10Mb browser error
00:51 sri umm yea, 5mb is the default limit
00:51 sri i thought we established that by now
00:52 vervain Ya... what I mean is 5MB that is too big.
00:52 sri see above for explanation of error
00:52 vervain So it's just above the limit... and you can get the 200 with error message.
00:52 sri you apparently can't
00:53 vervain But ya can. ;-)
00:53 sri safari is smart enough, the other browsers only get scared of the bus error
00:53 sri umm
00:53 sri broken pipe
00:53 purl broken pipe is probably a normal error, and memory usage could just be llvm being funny
00:53 vervain To create my test files I did 'dd if=/dev/urandom of=filex count=n'
00:53 sri anyway, my investigation is finished, patches welcome
00:54 vervain The 5MB trips the MAX_MESSAGE_SIZE error and the browsers display the correct 'File is too big.' message... the 10MB and the browser breaks.
00:55 sri yes, because it doesn't try to write to a closed socket
00:56 sri apparently those browsers don't like being interrupted
00:56 japhb Hmmm, Mojo::UserAgent looks useful for some network code we'd like to get more finer-grained parallelism out of.  OOC, is there anything that LWP::UA and friends do that Mojo::UserAgent currently does not?
00:57 japhb Meaning, did you design it to be a complete replacement?
00:57 sri i'm sure there is stuff, just like we have stuff LWP doesn't do
00:58 japhb OK, but I take it there wasn't a systemic effort to be a strict superset?
00:58 sri right
00:58 japhb OK, gotcha
00:59 sri design goal was to make web development as fun as possible without any restrictions
00:59 japhb I'm all about -Ofun.  ;-)
01:00 crab japhb: fwiw, i haven't found anything i couldn't do with mojo's client
01:00 japhb crab: Excellent, thank you.
01:02 sri japhb: since you're a perl6 fan, another goal is to have mojolicious ported entirely at some point
01:02 vervain I just did some more test with chrome... at 6, 7 and 8 MB it worked as expected (correct 'File is too big.' message') at 9MB it 'broke'.
01:02 japhb sri, I had seen something of the sort.
01:03 * japhb can't remember having mentioned Perl 6, huh.
01:03 sri vervain: lucky timing, it's prolly more or less random
01:03 japhb I know that someone started a Dancer port, but to say it is a minimal start at the moment is an understatement.
01:03 sri depending on chunk size and kernel send buffer size
01:03 vervain My experiments so far suggest a threshold... not random.
01:04 japhb And people occasionally mention something like "What Perl 6 really needs is a Mojolicious port"
01:04 vervain Hmm... maybe... I'll write some scripts to run tomorrow to get lots-o-data
01:04 sri imo what Perl6 needs first is perl6doc
01:05 vervain sri: the little people just found and broke the pickle :-(
01:05 elb0w joined #mojo
01:06 japhb There was work done on POD6 and introspectable doc support ... but where you meaning that you wanted the docs to be filled in, such that 'perl6doc -f foo' would give you what you want?
01:06 sri that's exactly what i want
01:07 sri don't care too much about tutorials, but reference documentation
01:07 japhb Nodnod
01:08 japhb Basically I think most of the people who have the collective knowledge to do that are scared off by the sheer enormity of it.
01:08 japhb I've heard things along the lines of wanting a Perl 6-friendly tchrist-equivalent to go off in a cave for 3 months and come back with docs.
01:09 sri i really don't think that's how it works
01:09 * japhb chuckles
01:09 sri one method at a time!
01:10 japhb Mostly people are throwing their effort there into working on the setting spec docs.
01:10 japhb Eventually those will need to be converted/updated to perl6doc-friendly versions
01:11 * japhb suddenly realizes what must have outed him as a Perl 6 fan -- mentioning -Ofun.  ;-)
01:12 sri actually i've seen you in #perl6
01:13 japhb Ah, well then, I suppose that's pretty damning evidence.
01:13 * sri nods
01:14 * japhb finds sadly that bringing it up as often as not gets in the way when talking to people who actively work on Perl 5 stuff.
01:17 sri well, it's a tricky topic
01:18 sri at this point i simply with the communities could be merged again
01:18 sri *wish
01:18 sri right now it's split into two extremes that aren't good for anyone
01:19 japhb I suspect there will be a core set of people that exist in both, and a (hopefully) smaller set that insist on only one.
01:19 japhb And I agree with that sentiment, though I've seen it in worse shape than now.
01:20 * japhb remembers when the verbal equivalent of Molotov cocktails were being thrown daily.
01:20 sri i've been watching it from the start
01:21 japhb So OOC, given that Mojo* is so big ... do you have any estimate of the effort that it will take to port it to Perl 6?
01:22 japhb (Ignoring for now the limitations of particular Perl 6 implementations, just under the assumption that they worked according to spec)
01:22 sri if everything works it actually shouldn't be too hard
01:23 sri mojolicious is just 11k lines of code
01:24 japhb Are the pieces at the very bottom layers useful by themselves, such that it can be built brick by brick and each brick brings something new and useful?  Or do you need to get a lot laid down before any of it is all that useful?  (My impression so far is the former, but I might not be aware of low-level ties between all the pieces)
01:24 sri even with all of its features
01:24 sri yes, it's a strickt brick by brick design
01:24 sri *-k
01:24 japhb Hmmmm
01:25 japhb (That's a good Hmmmm, by the way)
01:25 japhb afk for a bit to ruminate on this and eat dinner.  :-)
01:26 sri i'll prolly go to sleep
01:26 japhb Have a good night, then.
01:26 sri nn
01:27 crab doesn't the ruminating come AFTER eating?
01:55 sri vervain: found a great post about the topic (the last few sections) http://wiki.pylonshq.com/display/pylonscookbook​/Hacking+Pylons+for+handling+large+file+upload
01:55 sri sums up the connection reset problem pretty well
02:02 sri imo we handle the whole thing perfectly in mojolicious
02:03 sri it's strange though that modern browsers still behave so stupid
02:03 sri safari++
02:05 sri i mean google designs their own faster protocol (SPDY), but still can't handle interrupted uploads in chrome
02:20 japhb sri, if you're still around: Is there a doc or image that shows all the pieces of the whole Mojo* stack and how they fit together?
02:40 amoore joined #mojo
03:18 gshank joined #mojo
03:21 tempire japhb: there is not
03:21 tempire it's not that complicated, though
03:22 tempire it's mostly events being thrown using a loop.  an ioloop, even
03:30 gshank joined #mojo
03:35 japhb tempire, I was thinking in terms of the layers of bricks, so that I could see what things would need to be ported first.  :-)
04:12 heytrav joined #mojo
06:11 arpadszasz joined #mojo
06:19 Vandal joined #mojo
07:45 Foxcool joined #mojo
08:24 kaare joined #mojo
10:13 batman joined #mojo
10:23 crab sri: that pylons page is pretty horrifying
10:24 crab the page seems to be a long succession of "well, the last section sucked, but oh look, you can pile another hack on top to make it better" sections.
10:59 preflex_ joined #mojo
11:14 augensalat Hi guys! If my Mojolicious (REST) app has a resource for a certain HTTP method, it returns with a 404 for all other methods on that resource.
11:14 augensalat Is there an easy way to turn that into a more appropriate 405?
11:42 crab not without writing code for it, no.
11:43 crab it's conceivable that a plugin could be written to make it work, but it would probably have to do some dirty grovelling-around.
11:45 augensalat does that mean a patch for mojolicious?
11:45 augensalat Maybe there is a goot reason for that 404 response, that I'm not aware of...
11:48 sri why would you want a 405 response?
11:49 augensalat first of all it is confusing to get a "not found" for an existing resource.
11:49 augensalat then RFC 2616 has a quite clear meaning for that
11:50 augensalat §5.1.1 Method
11:53 sri if corretness is the only reason then i don't think it trumps maintainability
11:56 augensalat hmm ... I don't get the point
11:56 sri implementing it in core would be a total hack that makes the router less maintainable
11:57 sri so i agree with crab, it should be a plugin
11:58 augensalat ok then, I will try it that way
12:13 Psyche^ joined #mojo
12:27 ladnaV joined #mojo
12:39 batman joined #mojo
12:40 smpb joined #mojo
13:27 crab augensalat: rather than trying to make a generic plugin to do it, i would suggest a routing shortcut that creates a route with the desired method and a catch-other-methods route that returns 405.
13:29 crab it's probably easier than trying to decide in e.g. a before_dispatch hook whether a route exists for the current method or not.
13:32 crab or just don't worry about it. http has all sorts of "oh, wouldn't that be nice" sort of things that nobody actually uses or supports.
13:33 crab i can't easily imagine a situation in which a rest-client could do something useful to cope with a 405 error rather than a 404 one.
13:33 crab (but that's just my personal take on the matter. return 405 if you really care, by all means.)
13:36 crab may i also say that your nick is rather disturbing. makes me want to scrub my eyes.
13:43 Nickeeh joined #mojo
13:45 sri augensalat: in case crab didn't convince you yet, here's an example that should give you headaches "my $foo = $r->get('/foo'); $foo->websocket('/bar' => sub {…}); $foo->route('/:baz', baz => qr/\d+/)->to('foo#bar')"
13:46 sri routes are extremely dynamic, no matter how you implement it, 405 support will come with a hefty performance loss as well
13:49 sri in fact, if you really want more strict REST you might be better off building your own more restrictive little routing DSL with a plugin and shortcuts
13:51 andrefs joined #mojo
13:53 crab yes, to do it using routing shortcuts in specific cases would be much easier
14:39 MojoGuest22 joined #mojo
14:39 MojoGuest22 From: http://www.google.se/url?sa=t&amp;rct=j&amp;q=​irc%20client%20html5&amp;source=web&amp;cd=4&a​mp;sqi=2&amp;ved=0CDgQFjAD&amp;url=http%3A%2F%​2Fdev.xantus.org%2F&amp;ei=fzTaTvGrBMbb4QSgq7z​hDQ&amp;usg=AFQjCNGw6nZv0wO42uzUnHhWcD0vcKNULg (1 hits)
15:14 lammel2 joined #mojo
15:28 mire joined #mojo
15:30 gshank joined #mojo
15:32 jnap joined #mojo
16:13 mire joined #mojo
16:37 vervain joined #mojo
16:41 j3nnn1 joined #mojo
16:49 j3nnn1 left #mojo
16:56 noganex joined #mojo
17:07 xaka joined #mojo
17:09 marcus it makes rest clients easier to debug, to be sure.
17:09 marcus (supporting 405)
17:10 sri marcus: not worth the cost
17:12 ispy_ joined #mojo
17:30 ispy_ joined #mojo
17:30 jnap joined #mojo
17:43 andrefs joined #mojo
17:57 gshank joined #mojo
18:19 gshank joined #mojo
18:34 batman joined #mojo
18:47 gbacon joined #mojo
18:52 lammel2 joined #mojo
18:57 jnap joined #mojo
19:14 mire joined #mojo
19:16 gshank joined #mojo
19:30 andrefs joined #mojo
19:45 vervain joined #mojo
19:48 gshank joined #mojo
20:02 jnap joined #mojo
20:09 dmn001 joined #mojo
20:09 jnap joined #mojo
20:33 marcus sri: as a general thing I agree. When implementing a rest service it might.
20:34 sri marcus: please elaborate
20:35 perlite joined #mojo
20:36 sri what possible implementation have you evaluated to get to that conclusion?
20:43 marcus sri: I'm saying it might be useful to implement it as a shortcut when you're writing a web service
20:43 marcus sri: not that it would make sense to put it into the routes
20:43 sri ah
20:49 sri actually strange we don't have more shortcut plugins yet
20:52 MojoGuest419 joined #mojo
20:52 MojoGuest419 From: http://news.ycombinator.com/item?id=1277067 (166 hits)
20:53 marcus sri: I'm working on something, not sure if it's useful : http://nordaaker.no/docs/skitc​h/Untitled-20111203-215309.jpg
20:54 sri marcus: interesting start
20:54 vervain Very useful
20:54 sri want to show relationships?
20:54 MojoGuest419 is that irc not using any server thingz?
20:54 marcus sri: mostly groupings of stuff and where they belong in the layers.
20:54 marcus MojoGuest419: it is using a server thingz
20:55 marcus MojoGuest419: it communicates with the server thingz through web sockets.
20:55 sri thingz yo
20:55 MojoGuest419 what exactly?
20:55 marcus MojoGuest419: mojolicio.us
20:55 MojoGuest419 server is perl?
20:55 MojoGuest419 ax, ok :)
20:56 vervain irc.perl.org
20:56 purl rumour has it irc.perl.org is a round robin
20:58 sri marcus: not sure the layers make sense like that
20:59 sri i would group it more like (Mojo::Base, Mojo::Util, Mojo::EventEmitter…), (Mojo::IOWatcher, Mojo::IOLoop::Stream, Mojo::IOLoop), (Mojo::Path, Mojo::URL), (Mojo::JSON, Mojo::DOM, Mojo::UserAgent)
21:00 sri it's closer to their real relationships though
21:01 marcus sri: I tried putting all the things I considered 'utility' in the right column
21:01 marcus and the content generator/parsers together (mojo::dom/json/template)
21:01 sri i would put base and eventemitter in there too
21:02 sri (right)
21:02 marcus sri: everything uses base tho
21:02 sri about as utility as it gets :)
21:02 marcus sri: the eventemitter seems only related to the ioloop and everything under that tree?
21:03 sri nono, eventemitter is used all over the place for event emitting objects
21:03 sri Mojo::Log emits events ;p
21:03 marcus oh, I see it now
21:03 marcus and Mojo::Transaction
21:03 sri maybe one block for HTTP stack modules
21:03 sri transaction, message, path, params...
21:04 sri don't think useragent belongs in the io stack
21:04 sri it connects the io stack with the http one
21:04 marcus yeah, that's why they are on the left hand side together, with similar colors
21:05 marcus I think I should put all the 'utility stuff' in the bottom
21:05 marcus as you suggest
21:05 * sri wonders if that map could be done with overlapping circles
21:05 sri http://www.playframework.org/2.0 # like this :)
21:05 marcus sri: yeah, visualizing these things is quite tricky :)
21:06 sri yeah
21:07 sri iowatcher is the layer below ioloop::* btw. which is the layer below ioloop
21:07 elb0w joined #mojo
21:09 sri so Mojo::UserAgent is closed to Mojo::IOLoop than Mojo::IOLoop::Client
21:09 sri *closer
21:09 sri actually, it looks all a little mixed up :)
21:10 sri like asset above message
21:11 sri but i think this could be very useful
21:11 marcus well, I'll keep working on it.
21:12 marcus I think the fact that *I* am messing this up means it is useful ;-)
21:13 marcus we're watching revenge btw.
21:14 marcus really not sure where Mojo::Home belongs at all
21:14 sri utility seems the right category
21:14 marcus Kinda feels like it should actually be a Mojolicious class
21:15 sri home and cache
21:15 sri yea
21:15 sri home is used by Mojo though
21:16 sri command would also be a cnadidate
21:17 sri but in the end they are lower level framework building blocks
21:18 vel joined #mojo
21:20 marcus yeah.
22:02 marcus http://nordaaker.no/docs/skitc​h/Untitled-20111203-230209.jpg not sure if I'm making it better or worse, but now it fits on a A4 page =)
22:03 sri better :)
22:04 sri you forgot some though
22:05 marcus I'm making place for content::single atm
22:05 * sri sees Mojo::Template twice… could be the beer though
22:07 marcus no. that's fixed already tho
22:07 sri heh, where's my beer?
22:11 marcus i haz no beer.
22:11 marcus or rather, I have about 10 liters in the trunk of my car
22:11 marcus but none in my fridge
22:12 marcus actually more like 11 liters of strong christmas beer
22:12 marcus for the beer tasting at the nordaaker christmas party on tuesday
22:12 sri :o
22:13 marcus Hmm, missing Mojo::Loader
22:14 marcus guess it's an utility.
22:22 marcus http://nordaaker.no/docs/skitch/Mojolici​ous_Classes.graffle-20111203-232220.jpg
22:22 marcus I'm not going to include Command::* or Server::*
22:23 marcus missed anything else important? Still need to move things a bit about and change the colors and fix alignments.
22:23 sri right side doesn't look good imo
22:24 sri delay is very out of place
22:24 sri but i suppose your group only by color?
22:25 marcus yeah, and try to keep groups with relationship near eachother
22:25 marcus and I try to keep the most highlevel stuff near the top
22:25 sri hmm, maybe the background needs color too
22:26 marcus pink_
22:26 sri i like the color grouping
22:27 marcus I moved the empty hole to be where delay was
22:27 sri btw. morbo and hypnotoad might look good in there
22:28 sri and just get rid of Mojo::Server
22:28 sri some base classes like Mojo::Asset are also rather irrelevant
22:28 sri maybe remove all base classes that don't matter
22:29 sri message, content, cookie, transaction...
22:29 marcus yeah
22:30 sri btw. you could put Mojo::ByteStream above Mojo::Util, since they also have a special relationship
22:31 marcus or at least next to it
22:31 sri well, if a utility class has to be above another it would be bytestream
22:32 sri Mojo::Base and Mojo::EventEmitter also have a special relationship, the rest is just boring
22:33 sri but yea, it works rather well when you concentrate on the colors instead of order
22:38 marcus I actually did those changes as well
22:38 marcus http://nordaaker.no/docs/skitch/Mojolici​ous_Classes.graffle-20111203-233853.jpg
22:39 marcus this is where I'm at the moment
22:39 purl at the moment is just goes BZZZZZZZT if you do anything slightly wrong or learning how to use it
22:39 sri oh, i like this one the best
22:39 sri maybe morbo below hypnotoad
22:40 sri so ioloop client and useragent are closer
22:40 marcus am thinking transaction should maybe be below message
22:40 sri well, it connects messages and useragent/server
22:42 marcus yeah, from that perspective it has a good position
22:42 marcus I guess it's just as high level as messages
22:42 sri looks good to me otherwise
22:43 tempire I like that Mojo::CookieJar looks mangled
22:43 marcus http://nordaaker.no/docs/skitch/Mojolici​ous_Classes.graffle-20111203-234347.jpg
22:44 marcus did the morbo change as well, will start with alignment work now
22:44 marcus to get things to look a bit less messy :)
22:44 sri stream above delay
22:45 sri and ioloop next to stream
22:45 sri no wait… then theres a gap between ioloop and iowatcher
22:45 sri hmm
22:45 sri oh
22:46 sri delay to the right, ioloop to the left
22:47 sri actually, switch delay with ioloop and then ioloop with stream :D
22:47 marcus uuhm
22:47 marcus http://nordaaker.no/docs/skitch/Mojolici​ous_Classes.graffle-20111203-234722.jpg
22:47 marcus not like this then
22:47 sri no
22:47 sri almost
22:47 sri stream to the left
22:48 marcus yeah, that makes sense
22:48 sri \o/
22:48 marcus I guess this will be handy to print out as well
22:49 sri indeed
22:49 sri shouldn't ojo have its own color?
22:49 sri something darker
22:50 sri maybe a rainbow ;p
22:51 marcus and a unicorn!
22:51 purl hmmm... a unicorn is a reference to the django pony
22:51 * tempire glitters
22:52 sri tempire: you might want to see a doctor with that
22:54 * tempire bedazzles sri
23:03 marcus http://nordaaker.no/docs/skitch/Mojolici​ous_Classes.graffle-20111204-000253.jpg
23:03 marcus I'm starting to like it now.
23:03 * sri too
23:05 sri maybe transactor could be moved a bit
23:06 sri so it's not between client and useragent
23:07 sri not sure if to the left or up though
23:08 sri would be nice if transactor touched the transactions
23:09 marcus "Tell me where the bad transaction touched you"
23:17 marcus sri: you made me move everything around again :) http://nordaaker.no/docs/skitch/Mojolici​ous_Classes.graffle-20111204-001743.jpg
23:18 sri marcus: upload and message need to swap places ;p
23:19 sri headers are actually closer to content than message :o
23:21 marcus http://nordaaker.no/docs/skitch/Mojolici​ous_Classes.graffle-20111204-002047.jpg like so?
23:21 sri yea
23:22 sri maybe move headers and date to the right?
23:22 sri cookiejar to the top above the cookies :)
23:22 sri and then i think we have it
23:23 sri OR
23:23 sri switch cookiejar and url
23:29 marcus http://nordaaker.no/docs/skitch/Mojolici​ous_Classes.graffle-20111204-002047.jpg
23:29 marcus I also rounded the corners a bit and reduced the drop shadow
23:30 Patterner will the .graffle file be available?
23:31 marcus It can be.
23:32 marcus I'll just stick it on github, and you buggers can update it when sri adds more classes :D
23:39 marcus wonder how hard it would be to bridge mojolicious and coffeescript
23:49 jnap joined #mojo
23:51 marcus hay jnap
23:51 jnap hey!
23:57 marcus I haz blogged - http://marcus.nordaaker.com/2011/12/v​isualization-of-the-mojo-class-tree/
23:58 sri marcus: you sure you want to blog that on a weekend?

| Channels | #mojo index | Today | | Search | Google Search | Plain-Text | summary