#rest on irc.freenode.net 2011-12-19 (txt)

[00:02] kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Quit: Computer has gone to sleep.
[00:04] kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #rest.
[00:11] Hakon|mbp (~hakon1@193.137.202.84.customer.cdi.no) joined #rest.
[00:12] mogsie (~mogsie@62.101.198.35) left irc: Quit: Leaving.
[00:18] mogsie (~mogsie@62.101.198.35) joined #rest.
[00:23] Hakon|mbp (~hakon1@193.137.202.84.customer.cdi.no) left irc: Quit: Leaving...
[00:24] kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Quit: Computer has gone to sleep.
[00:40] nuclearsandwich (~nuclearsa@adsl-64-175-40-88.dsl.pltn13.pacbell.net) left irc: Remote host closed the connection
[00:42] <mikekkkk> :)
[00:44] Hakon|mbp (~hakon1@sfn-inkubator-70-72.hib.no) joined #rest.
[01:00] ssedano (~ssedano@unaffiliated/ssedano) joined #rest.
[01:34] kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #rest.
[01:38] mr_yall -> mryall`
[01:38] mryall` (~mryall@scifo.mexico86.co.uk) left #rest ("Killed buffer").
[01:41] pc1oad1etter (~pc1oad1et@ec2-174-129-99-58.compute-1.amazonaws.com) left irc: Ping timeout: 252 seconds
[02:00] mephju (~mephju@dslb-094-222-015-193.pools.arcor-ip.net) joined #rest.
[02:06] <mogsie> but mikekkkk, You brushed off my questions as "application specific" and that HTTP shouldn't meddle in this, but nonetheless, an application designer will be faced with these questions, and anyone implementing a _generic UA_ would also wonder about these questions.
[02:20] <mikekkkk> I have no idea what a generic UA is
[02:20] <mikekkkk> if you mean an HTTP client then it will know everything useful it needs to know
[02:21] <mikekkkk> i.e. the request is idempotent
[02:21] <mikekkkk> I've already asked several times for an explanation of what infrastructure does/could/will rely on the fullness of PUT
[02:21] <mikekkkk> and I haven't been provided with one single example.
[02:21] <mikekkkk> go figure.
[02:22] <mikekkkk> every time we get ot that point
[02:22] <mikekkkk> people stop responding.
[02:23] <mikekkkk> and/or revert back to some argument like "well that's just how PUT is specified now, go look at httpbis changes"
[02:23] <mikekkkk> which is not a logical argument, it's just a way of avoiding an unanswerable question
[02:25] <mikekkkk> I could point to at least 3 or 4 branches of that thread that stopped due to me asking for that example or even just some basic rationale
[02:27] <mikekkkk> that Fielding response is a joke or waht? :/
[02:37] <mogsie> A Generic UA is like a browser.
[02:37] <mikekkkk> right
[02:37] <mogsie> or any other type of client that doesn't have any out-of-band information (<evil grin>)
[02:37] <mikekkkk> the majority of its behaviour is specified by HTML
[02:37] <mikekkkk> not the semantics of HTTP
[02:38] <mikekkkk> a browser cache is *infrastructure*
[02:38] <mikekkkk> it's an intermediary
[02:38] <mogsie> I agree with you that there aren't any good arguments as to why the fullness "constraint" / "suggestion" is there in 2616.
[02:38] <mikekkkk> oh.
[02:39] <mikekkkk> well ok then that is my only point
[02:39] <mikekkkk> :)
[02:39] <mogsie> :-)
[02:39] <mikekkkk> the problem here is Roy's further dug a hole
[02:39] <mikekkkk> for no apparent reason
[02:39] <mogsie> I would say it's a gray area.
[02:40] <mikekkkk> and now we have people (like rails guys) quoting him and using his opinion as justification for implementing features against Roy's not-yet-published theory on how PUt should work
[02:40] <mogsie> I do like the PUT is not partial approach of doing things.
[02:40] <mogsie> I don't have to mint yet another media type to describe arbitrary differences in my other media type.
[02:41] <mikekkkk> but you're minting either media types of link relations anyway..
[02:41] <mogsie> GET foo, edit it, PUT foo. It's a very well understood interaction.
[02:41] <mikekkkk> so you're already communicating application semantics to your clients..
[02:42] <mogsie> Of course, but that's not out-of-band, it's part of the "common knowledge" of the media types/rels, as you put it.
[02:42] <mikekkkk> the semantic 'my application only permits complete replace behaviour with PUT requests' is not hard to document
[02:42] <mogsie> Atompub say to do exactly this.
[02:42] <mikekkkk> mogsie: the reason someone is not out of band
[02:42] <mikekkkk> is so that it's self descriptive and visible in the interaction
[02:42] <mikekkkk> the reason you want something to be SD and visible is so that intermediaries can understand it and build mechanisms around it
[02:43] <mikekkkk> nobody is doing that with thef ullness of PUT
[02:43] <mikekkkk> no-one
[02:43] <mikekkkk> 0.
[02:43] <mikekkkk> :|
[02:43] <mikekkkk> so the fact that the fullness is out of band
[02:43] <mikekkkk> is basically irrelevant - it's not a negative because it was never used anyway
[02:44] <mikekkkk> this is why I keep asking for examples of intermediary mechanisms
[02:44] <mikekkkk> because I understand what the argument about 'out of band' is about.. I just don't believe it is a valid argument in this particular case
[02:44] mr_yall (~mryall@scifo.mexico86.co.uk) joined #rest.
[02:44] <mikekkkk> if you took the safety semantic off GET
[02:44] <mikekkkk> that would fuck up a lot of stuff
[02:45] <mikekkkk> same with pretty much every semantic property of all the Methods.. apart from this full replace semantic of PUT
[02:45] <mikekkkk> it's basically pointless and it prevents certain type of interaction that are demonstrably useful
[02:46] <mikekkkk> but hey - if Roy specifies it, it must be necessary
[02:46] <mogsie> heh
[02:46] <mikekkkk> heil Web Jesus
[02:46] <mikekkkk> :|
[02:46] <mogsie> Luckily, the Thesis doesn't mention any methods, only the uniform interface.
[02:46] <mikekkkk> indeed!
[02:46] <mikekkkk> I think Roy has a lot of sway in HTTP as well though
[02:47] <mikekkkk> so I don't think I'll go there.
[02:47] <mikekkkk> :)
[02:47] <mogsie> If an origin server decides to support partial updates, then nothing in the web will break, since it would have to support the Idempotency anyway.
[02:47] <mikekkkk> right
[02:47] <mikekkkk> that is happening all the time with rails already
[02:47] <mogsie> It's an idempotent unsafe method, That's what concerns the infrastructure.
[02:47] <mikekkkk> yup +1
[02:48] <mogsie> If I send in a JSON field with 2Mb of whitespace in it, that whitespace will be ignored; it's not stored anywhere.
[02:48] <mogsie> Likewise, an XML document with loads of  in it will also likely be ignored.
[02:48] <mikekkkk> right - that's more to do with json's processing model
[02:48] <mikekkkk> but yeah
[02:49] <mikekkkk> the stupid thing is you can hack around the partial PUT thing anyway
[02:49] <mogsie> They're "less semantically significant"
[02:49] <mikekkkk> by just using a different media type for each partial at the same URI
[02:49] <mogsie> Mm
[02:49] <mikekkkk> and then technically you aren't breaking anything
[02:49] <mikekkkk> it's stupid
[02:49] <mikekkkk> and I would never do it
[02:49] <mikekkkk> but that's a good demo of how dumb the semantic is
[02:50] <mogsie> Well, while I agree with you that partial PUT isn't evil, I don't really see the benefit of it either...
[02:50] <mogsie> Would it be to save bandwitgh?
[02:50] <mikekkkk> application/partial+foo,bar,boo+json
[02:50] <mogsie> ugh width...
[02:50] <mikekkkk> bandwidth and deal with shitty networks
[02:50] <mikekkkk> like mobile networks
[02:50] <mogsie> ah, yes.
[02:50] <mogsie> that
[02:50] <mikekkkk> that's getting more common now right?
[02:50] <mikekkkk> :S
[02:51] <mogsie> Why is using POST with partials (like the web has been running for 10+ years now) using application/www-form-urlencoded-thingy
[02:52] <mogsie> if I POST "username=foobar&name=Foo" then that's a non-idempotent partial update. What's wrong with that?
[02:52] <mikekkkk> nothing, personally I think patch is kind of pointless
[02:52] <mikekkkk> PATCH^
[02:52] <mogsie> +1 :-)
[02:53] <mikekkkk> the requirement for PATCH was basically born out of this desire for PUT to not allow partials
[02:53] <mikekkkk> and then they were like
[02:53] <mikekkkk> "oh shit now we only have POST and no partial capability at all"
[02:53] <mikekkkk> "i know! let's invent a partial method, problem solved"
[02:53] <mogsie> A generic UA or any intermediary won't _really_ know that the POST is handled in an idempotent manner by the origin server, but losing out on that is pointless.
[02:53] <mogsie> But code-on-demand could fix that by passing knowledge to the client at run-time.
[02:54] <mogsie> Which is how the whole "web 2.0" works, right?
[02:54] <mikekkkk> I guess
[02:54] <mikekkkk> fwiw rails defaults to hidden field named _method
[02:54] <mikekkkk> and tunnels html forms all through POST
[02:55] <mikekkkk> but you can still hit a rails API 'properly' w/ PUT
[02:55] <mikekkkk> but the html stuff doesnt do that by default to deal with js-less browsers
[03:06] <mikekkkk> I'm super interested to read Roy's response my latest post
[03:06] <mikekkkk> either way it's going to clear up in my mind where he is coming from
[03:07] <mikekkkk> .. hopefully.
[03:07] <mikekkkk> :)
[03:07] mephju (~mephju@dslb-094-222-015-193.pools.arcor-ip.net) left irc: Quit: Ex-Chat
[04:25] Hakon|mbp (~hakon1@sfn-inkubator-70-72.hib.no) left irc: Ping timeout: 248 seconds
[04:46] cwb` (~user@russell.carlbaatz.com) joined #rest.
[04:46] blongden_ (~blongden@nocarrier.co.uk) joined #rest.
[04:46] fu-manch1 (~fumanchu@adsl-99-30-180-185.dsl.sfldmi.sbcglobal.net) joined #rest.
[04:46] KeeperOfTheSoul (~chris@94-193-221-187.zone7.bethere.co.uk) joined #rest.
[04:46] blongden (~blongden@nocarrier.co.uk) left irc: Ping timeout: 244 seconds
[04:46] chilversc (~chris@about/csharp/regular/KeeperOfTheSoul) left irc: Ping timeout: 244 seconds
[04:46] kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) left irc: Ping timeout: 244 seconds
[04:46] twilliams (~twilliams@apache/committer/twilliams) left irc: Ping timeout: 244 seconds
[04:46] trygvis (trygvela@knuth.ping.uio.no) left irc: Ping timeout: 244 seconds
[04:46] mamund (mamund@obsidian.nullshells.net) left irc: Ping timeout: 244 seconds
[04:46] impl (impl@atheme/member/impl) left irc: Ping timeout: 244 seconds
[04:46] KeeperOfTheSoul (~chris@94-193-221-187.zone7.bethere.co.uk) left irc: Changing host
[04:46] KeeperOfTheSoul (~chris@about/csharp/regular/KeeperOfTheSoul) joined #rest.
[04:47] Blazeix (~Blazeix@71.74.190.197) left irc: Ping timeout: 240 seconds
[04:47] axisys (~axisys@ip68-98-189-233.dc.dc.cox.net) left irc: Ping timeout: 240 seconds
[04:47] fu-manchu (~fumanchu@99.30.180.185) left irc: Ping timeout: 240 seconds
[04:47] cwb (~user@russell.carlbaatz.com) left irc: Ping timeout: 240 seconds
[04:47] lmarburger (u1979@gateway/web/irccloud.com/x-hslxyrgwuwggrrtw) left irc: Ping timeout: 240 seconds
[04:47] twilliams (~twilliams@apache/committer/twilliams) joined #rest.
[04:47] trygvis (trygvela@knuth.ping.uio.no) joined #rest.
[04:47] mamund_ (mamund@obsidian.nullshells.net) joined #rest.
[04:47] #rest: mode change '+o mamund_' by ChanServ!ChanServ@services.
[04:47] impl_ (impl@ester.at.cynigram.com) joined #rest.
[04:47] impl_ (impl@ester.at.cynigram.com) left irc: Changing host
[04:47] impl_ (impl@atheme/member/impl) joined #rest.
[04:47] axisys_ (~axisys@ip68-98-189-233.dc.dc.cox.net) joined #rest.
[04:47] mamund_ (mamund@obsidian.nullshells.net) left irc: K-Lined
[04:48] <mikekkkk> ewww
[04:48] blongden_ (~blongden@nocarrier.co.uk) left #rest.
[04:52] KarlHungus (~relax@unaffiliated/adj) got netsplit.
[04:52] KarlHungus (~relax@unaffiliated/adj) returned to #rest.
[04:55] lmarburger (u1979@gateway/web/irccloud.com/x-yaegcokcrmboepxi) joined #rest.
[04:57] KarlHungus (~relax@unaffiliated/adj) got netsplit.
[04:59] Blazeix (~Blazeix@71.74.190.197) joined #rest.
[05:00] KarlHungus (~relax@unaffiliated/adj) returned to #rest.
[05:10] mikekkkk 's spidey senses are suggesting I may be headed for some internet hand-bags with Roy
[05:25] <mikekkkk> I really want to meet Eric J Bowman in real life
[05:28] <mikekkkk> just to see if he's as confrontational in real life as he is online
[05:28] <mikekkkk> at least I'm consistently confrontational.
[05:29] <mikekkkk> I can't believe he just accused me of trolling
[05:29] <mikekkkk> fucking cheek!
[06:19] grove (~grove@aggw006.cappelendamm.no) left irc: Quit: grove
[06:30] <mogsie> Does anyone know anything about "wemi" at oasis?
[06:30] <mogsie> http://lists.oasis-open.org/archives/tc-announce/201112/msg00007.html
[06:30] <mogsie> http://www.oasis-open.org/committees/tc_home.php?wg_abbrev=wemi
[06:32] <mikekkkk> where is steve, does he not frequent here any more ?
[06:33] <mikekkkk> he's a cool dude but this is silly: https://github.com/rails/rails/pull/505#issuecomment-3203358
[06:44] <mikekkkk> someone reassure me I'm not just coming across like a troll
[06:44] <mikekkkk> :)
[06:45] <darrelmiller> In principle I agree with you.
[06:46] <mikekkkk> good enough for me
[06:46] <mikekkkk> no need to clarify
[06:46] <darrelmiller> I'm trying to write a response to Fielding at the moment. I think he hit the nail on the head when he said "PUT means PUT". There is this eimplicit understanding of what it means, but whenever they try and put it in writing the description is full of wholes.
[06:46] <mikekkkk> oh damn you did it anyway
[06:46] <mikekkkk> :(
[06:46] <darrelmiller> s/wholes/holes.
[06:47] <mikekkkk> the thing is (and it came out in Jan's response) - people think i'm contesting the intention
[06:47] <mikekkkk> I'm not contesting the intention
[06:47] <mikekkkk> I'm contesting how the spec is actually written
[06:47] <mikekkkk> and more importantly what the actual downsides are to relaxing it
[06:47] <darrelmiller> Well, I think your solution would make the spec easier to write clearly.
[06:47] <mikekkkk> indeed.
[06:48] <darrelmiller> and I agree that there are no negative side effects except for making a bunch of smart people look dumb.
[06:48] <mikekkkk> I dunno whether I just don't make enough sense when I write on that list
[06:48] <mikekkkk> becuase I seem to infuriate people a lot.
[06:48] <mikekkkk> and I do honestly want to get somewhere rather than go round in circles..
[06:48] mikekkkk shrugs
[06:49] <darrelmiller> They are infuriated because it is extremely difficult to defend the complete replace semantics argument because there are so many exceptions to it.
[06:49] <mikekkkk> I didn't even think the way I was writing was that grating
[06:49] <mikekkkk> I've got a good 6 gears to go yet as far as writign abrasively :P
[06:50] <darrelmiller> I don't think you are being grating other than the fact that you won't drop a difficult subject.
[06:50] <mikekkkk> but so far I get Roy beign snarky and condescending, Eric calling me a troll and trying to make me look like a dick, and Jan being short and talking to me like I'm an idiot
[06:50] <mikekkkk> and all I'm really doing is asking them questions
[06:52] <darrelmiller> If they would use a term like "conceptually replace the resource" then I could live with that. I don't have the same burning desire to be able to do partial updates via PUT.
[06:52] <mikekkkk> if asking people hard questions makes me a dick and a troll then that community is in pretty bad shape.
[06:53] <mikekkkk> if all else fails I'm going to bring up this "multiple media types for each individual possible partial" approach
[06:53] <mikekkkk> because that will /really/ fuck them up
[06:53] <mikekkkk> application/partial+a,b,c+xml
[06:53] <mikekkkk> oh look, it's a "full replace"
[06:54] <mikekkkk> but wait that's a partial for the resource
[06:54] <darrelmiller> huh? That's over my head.
[06:54] <mikekkkk> yet but we deal in represntation transfer
[06:54] <fu-manch1> of course Eric's reply will be "just use HTML for all of them" ;)
[06:54] <mikekkkk> I',m saying you could produce a media type for each partial
[06:54] <mikekkkk> for a resource
[06:54] <mikekkkk> and then "full PUT" a partial media type
[06:55] <mikekkkk> depending on what bit of the resource you wanted to update
[06:55] <darrelmiller> ok, but if you PUT the partial and then you GET you would get the partial, not an updated original resource.
[06:55] <mikekkkk> application/partial+name,age+json
[06:55] <mikekkkk> well if you full PUT that
[06:55] <mikekkkk> and then asked for application/json
[06:56] <darrelmiller> Surely, you would just get back the partial you PUT, no?
[06:56] <mikekkkk> no you asked for a diff media type
[06:56] <mikekkkk> diff representation of the same resource
[06:56] <darrelmiller> Ahhhh... representations as different subsets of the same resource.
[06:56] <darrelmiller> interesting.
[06:56] <mikekkkk> exactly
[06:57] <mikekkkk> you would never do that
[06:57] <mikekkkk> but that proves how stupid the complete replace semantic really is
[06:57] <darrelmiller> I didn't think that representation were really supposed to do that.
[06:57] <mikekkkk> I haven't brought it up yet cos it just complicates the issue
[06:58] <mikekkkk> well they aren't but theycan't really argue with GET'ing PUT'ing partials
[06:58] <darrelmiller> I know representation do contain different subsets of data but I thought that was supposed to be related to limitations of the format rather than selectively choosing different pieces of the resource.
[06:58] <mikekkkk> and GET'ing complets
[06:58] <mikekkkk> it's inherent in thew hole resource>representation relationship
[06:59] <mikekkkk> (and it's the opposite side of the reason why PUT is never used in practice to prime the cahce)
[07:03] <mikekkkk> darrelmiller: how do you actually determine that distinction
[07:03] <darrelmiller> between resource and representation?
[07:04] <mikekkkk> no between whether or not representations contain equivalent levels of information
[07:04] <mikekkkk> or subsets/etc ?
[07:04] <mikekkkk> this is why I haven't brought this up btw cos it's just distracting and weird :)
[07:05] <darrelmiller> Roy once said something like, "representations should be mechanical translations of a resource".
[07:05] <mikekkkk> should
[07:05] <mikekkkk> :)
[07:05] <darrelmiller> I try to ensure that there is no "logic" involved when creating a respresenation from a resource.
[07:05] <mikekkkk> wel all know what that means.. ;)
[07:06] <mikekkkk> darrelmiller: yeah me too but it's hard to get around the fact that a resource could be implemented that way
[07:06] <mikekkkk> and techncially it wouldn't violate HTTPbis or REST
[07:06] <mikekkkk> so..
[07:06] <mikekkkk> :|
[07:06] <mikekkkk> again, I'm not saying that it's a good idea or I would ever do it
[07:06] <darrelmiller> You might not gain any more credibility with that argument :-\
[07:06] <darrelmiller> however, correct you are.
[07:06] <mikekkkk> right that's why I haven't brought it up
[07:07] <mikekkkk> :)
[07:07] <mikekkkk> I would be interested to hear how Roy ties that up though
[07:07] <mikekkkk> still waiting on the last post though..
[07:07] <mikekkkk> :)
[07:08] <darrelmiller> The point I'm trying to convey is that telling new people that you can't do partial updates is very confusing when you then say, but you don't need to send links, or read only fields or "last updated" fields, etc.
[07:09] <darrelmiller> There are so many exceptions to this "complete replacement" rule that is is extremely difficult for people to know what is the right thing to do.
[07:10] <fu-manch1> "complete but perhaps slightly malformed or in need of transformation update" just doesn't have the same ring to it
[07:10] <mikekkkk> right but the answer to that from their pov is..
[07:10] <mikekkkk> use diff media types up and down
[07:10] <mikekkkk> ala html and form-encoded
[07:10] <darrelmiller> you could also do PUT /Customer/99?Includes="name,age,weight" and the side-effect of is to do a partial update on /Customer
[07:10] <mikekkkk> yeah I fucking hate those APIs
[07:10] <darrelmiller> LOL
[07:10] <mikekkkk> seriously - they annoy the shit out of me
[07:11] <mikekkkk> just PUT partials.. ffs
[07:11] <mikekkkk> -_-
[07:11] <darrelmiller> fu-manch1: "partially complete update" :-)
[07:11] <fu-manch1> :D
[07:11] fu-manch1 -> fumanchu
[07:11] <mikekkkk> ha
[07:11] <mikekkkk> yeah I bet the resposne to that is 'use diff media type up and down' or spec your media type so it's diff up than down
[07:12] <darrelmiller> "complete replacement of only the important stuff "
[07:12] <mikekkkk> which is when I will be tempted to bring up my super crazy media-types-for-every-possible-partial approach
[07:13] <fumanchu> yes, my Shoji media type basically says "there's a {body: {}} member in the JSON. Everything in there is the "important stuff" and everything outside of it is only "important" on insert, not update"
[07:13] <darrelmiller> mikekkk: To say you should do a complete replacement and use form-enc up and HTML down is going to be f*cking confusing 99% of devs.
[07:13] <mikekkkk> right, I know that you're preachign to the converted
[07:14] <mikekkkk> :P
[07:14] <darrelmiller> fumanchu: Nice.
[07:14] <mikekkkk> the whole thing is fucking insane as far as I'm concerned
[07:14] grove (~grove@84.115.45.31.customer.cdi.no) joined #rest.
[07:14] <mikekkkk> afaict, this comes down to a bunch of people in control of a spec who want it to mean something they originally intended
[07:14] <mikekkkk> fo rthe sake of it meaning what the intended
[07:14] <mikekkkk> rahter than because it's useful
[07:15] <mikekkkk> or important in some way
[07:15] <mikekkkk> which is a bullshit reason, frankly.
[07:15] <mikekkkk> what they intended is basically irrelevant
[07:16] <mikekkkk> if itmattered that much they should've spec'd it right in the first place
[07:18] bradley-holt (~bradley-h@65-183-135-35-dhcp.burlingtontelecom.net) joined #rest.
[07:18] <mikekkkk> I think it' pretty interesting that a lot of APIs that appear to use replace PUT
[07:18] <mikekkkk> actually *explicitly* specify in their own way - that their PUTs are full
[07:19] <mikekkkk> if HTTP was cut and dry.. why the hell would they need to do that?
[07:20] <mikekkkk> and - more importantly - their application will suffer nothing at all were PUT to be 'relaxed'
[07:22] <mikekkkk> sorry I'll shut up now
[07:22] steveklabnik (~steveklab@c-67-165-87-244.hsd1.pa.comcast.net) joined #rest.
[07:22] <steveklabnik> mikekkkk: hey!
[07:23] <steveklabnik> i've been off of irc for the last week or so because I'm trying to Get Shit Done, specifically with my 'book'
[07:23] <steveklabnik> but I do read the backlogs to this channel on the web, so I can keep up
[07:23] <steveklabnik> and I saw you mention me. sooo
[07:23] <steveklabnik> <mikekkkk> someone reassure me I'm not just coming across like a troll
[07:23] <steveklabnik> to be honest, I think you are.
[07:23] <steveklabnik> I dont think it's intentional
[07:24] <steveklabnik> i think the big deal is that i (and others) are saying 'let's implement the spec
[07:24] <steveklabnik> and you're saying 'the spec is stuipd'
[07:24] <steveklabnik> which are two separate things.
[07:24] <darrelmiller> steveklabnik: He's not a troll, he's just a pitbull. Just can't let go.
[07:24] <steveklabnik> the fact that infrastructure may or may not use PUT's semantics are irrelevant to me.
[07:24] <steveklabnik> i care what the spec says.
[07:24] <steveklabnik> darrelmiller: heh.
[07:25] <steveklabnik> also, it went from 'maybe troll, but i'll give him the benefit of the doubt' to 'omg what a troll' once you start ARGUING WITH THE GUY WHO WROTE THE SPEC OVER WAHT THE SPEC MEANS!!!
[07:25] <steveklabnik> ... but that's fine. and i say that with <3 in my tone and in my heart.
[07:25] <steveklabnik> but i think that's where the division comes from.
[07:25] <steveklabnik> They are infuriated because it is extremely difficult to defend the complete replace semantics argument because there are so many exceptions to it. <- I see this as the opposite
[07:26] <steveklabnik> partial whatever has sooo many edge cases
[07:26] <steveklabnik> complete replace is simple: store this.
[07:26] <steveklabnik> done.
[07:26] <steveklabnik> not complicated.
[07:26] <darrelmiller> But that's not practical, and not how people use it unless they are storing files.
[07:26] <steveklabnik> it is practical.
[07:27] <steveklabnik> people dont use it because they dont read the spec.
[07:27] <darrelmiller> So how do you do a PUT with a resource that has a last updated field?
[07:27] <steveklabnik> and they just Do Whatever The Framework Does.
[07:27] <steveklabnik> calculate one and PUT it?
[07:27] <steveklabnik> that's when you last updated it, after all.
[07:27] <darrelmiller> So each client uses it's own clock?
[07:27] <steveklabnik> you've got UTC offsets.
[07:28] <steveklabnik> also
[07:28] <darrelmiller> And do you really allow the client to PUT a resource that contains links that other clients might read and follow?
[07:28] <steveklabnik> one second, i'm still reading the rest of this backlog
[07:28] <steveklabnik> also, im on cup of coffee #1
[07:28] <darrelmiller> How does the client construct the URI?
[07:28] <darrelmiller> Ok, we'll talk once you are are on cup #3. :-)
[07:30] mamund (~mamund@rrcs-24-172-200-42.central.biz.rr.com) joined #rest.
[07:30] #rest: mode change '+o mamund' by ChanServ!ChanServ@services.
[07:30] <steveklabnik> anyway
[07:30] mamund is here
[07:30] <steveklabnik> okay, read the rest of the backlog
[07:30] <steveklabnik> mamund: zomg hi!
[07:30] <steveklabnik> check the chat logs, i just ... ranted. ish.
[07:30] <steveklabnik> anyway, darrelmiller
[07:30] mamund has been k-lined from his shell account!
[07:30] <darrelmiller> mamund: You might want to hide mikekkkk is on a roll.
[07:30] <mamund> ROFL
[07:31] <steveklabnik> i would suggest that not PUT-ing a 'last updated' field is just fine
[07:31] <mikekkkk> steveklabnik: I undestand exactly what you saying
[07:31] <steveklabnik> because resources are allowed to be translated by business rules
[07:31] <mikekkkk> I am clarifying something for you
[07:31] <mikekkkk> 2616 is NOT clear enough
[07:31] <darrelmiller> steveklabnik: but then it is not a complete replacement
[07:31] <mamund> i figured it was getting rough since mikekkkk was posting tao quotes to twitter
[07:31] <mikekkkk> so you are unnecessarily churning rails for *no reason*
[07:31] <steveklabnik> lol
[07:31] <mikekkkk> until httpbis pushes through
[07:31] <mikekkkk> you have no pressing need to churn rails
[07:31] <mikekkkk> period.
[07:31] <mikekkkk> you don't.
[07:31] <mikekkkk> fact.
[07:31] <mikekkkk> read 2616
[07:31] <mikekkkk> it's all over the place.
[07:31] <steveklabnik> well, while we present our facts as opinion
[07:31] <steveklabnik> I think it's perfectly clear
[07:31] <steveklabnik> so I do have a pressing need. period.
[07:32] <steveklabnik> ;)
[07:32] <mikekkkk> what is that?
[07:32] <mikekkkk> what isnt working?
[07:32] <mikekkkk> what's breaking?
[07:32] <mikekkkk> nothing.
[07:32] <steveklabnik> well, if we're going by 'working,' let's just ignore everything
[07:32] <mikekkkk> the only logcial argument I can understand is
[07:32] <steveklabnik> GET /resource?action=delete certainly _works_
[07:32] <mikekkkk> "I have OCD. this must be done"
[07:32] <steveklabnik> to a certain extent, that is true.
[07:32] <mikekkkk> well there you go.
[07:32] <steveklabnik> i feel that standards are very important.
[07:32] <steveklabnik> and should be followed.
[07:32] <mikekkkk> terrific
[07:32] <steveklabnik> that's the only way that things get done.
[07:32] <mikekkkk> you're going to make rails more complex
[07:33] <steveklabnik> i do not think they are _infallible_
[07:33] <mikekkkk> for th esake of satisfying some ocd urge
[07:33] <steveklabnik> at some level, isn't following any standard 'ocd'?
[07:33] <mikekkkk> that is pretty messed up, and you should probably be more transaprent about that in the thread
[07:33] <steveklabnik> i've already said this.
[07:33] <mikekkkk> er..
[07:33] <steveklabnik> standards are important to follow; that's why they're standards.
[07:33] <mikekkkk> no shit lol :)
[07:33] <mikekkkk> this just happens to be part of a standard that isn't clear
[07:33] <mikekkkk> and serves no logical purpose
[07:33] <steveklabnik> so why is your following a standard 'ocd' and mine isn't?
[07:33] <mikekkkk> or functional purpose
[07:33] <mikekkkk> so..
[07:34] <mikekkkk> if you're fine as you are
[07:34] <mikekkkk> I would wait until you absolutely HAVE to move
[07:34] <mikekkkk> which you don't
[07:34] <mikekkkk> until httpbis comes into play
[07:34] <mikekkkk> you have no real reason to change - 2616 is contentious which is *exactly* why it fot revised
[07:34] <mikekkkk> if it wasnt
[07:34] <mikekkkk> it wouldn't have been changed.
[07:34] <mikekkkk> how is that hard to accept?
[07:34] <steveklabnik> I think that's a perfectly fine statement.
[07:35] <mikekkkk> so.. what is the problem here?
[07:35] <steveklabnik> there is no problem.
[07:35] <mikekkkk> how am I trolling. I'm being sensible.
[07:35] <steveklabnik> except you're getting your panties in a twist over me wanting to add a configuration option over http verbs.
[07:35] <mikekkkk> IU'm not fucking about with shit for sake of satisfying OCD urges.
[07:35] <steveklabnik> right, see
[07:35] <mikekkkk> thati s not a valid argument
[07:35] <mikekkkk> it's a plea for help.
[07:35] <steveklabnik> that's why you come across as a troll
[07:35] <mikekkkk> there are other solutions to that plea
[07:35] <steveklabnik> your rhetoric is too harsh.
[07:35] <mikekkkk> like.. see a doctor.
[07:35] <mikekkkk> :)
[07:36] <mikekkkk> that's a joke btw
[07:36] <steveklabnik> arguing that people are mentally unstable is a _great way_ to bring them around to your side.
[07:36] <mikekkkk> lol cmon
[07:36] <steveklabnik> ;)
[07:36] <mikekkkk> seriously?
[07:36] <steveklabnik> i dunno
[07:36] <mikekkkk> that's not a very anarcho-anything pov
[07:36] <mikekkkk> :)
[07:36] <steveklabnik> the crux of your argument is that 'steve has ocd, leave it alone'
[07:36] <steveklabnik> ;)
[07:36] <mikekkkk> no the crux of my argument is
[07:36] <steveklabnik> (i am not actually upset)
[07:36] <mikekkkk> everything is fine, leave it you have no pressing requirement to revise anything
[07:37] <mikekkkk> all yuo will do is cause disruiption and nothing will be achieved
[07:37] <steveklabnik> i think everything is fine because rails tunnels over POST
[07:37] <steveklabnik> that's the real irony here.
[07:37] <mikekkkk> ...
[07:37] <steveklabnik> POST would map just fine to the rails update action.
[07:37] <mikekkkk> how is that ironic it's just weird.
[07:37] <mikekkkk> you're accepting it doesn't matter - then when I say it doesn't matter and nothing needs to be changed
[07:37] <mikekkkk> you tell me I'm being trollish..
[07:37] <mikekkkk> wtf
[07:37] <steveklabnik> being a troll is about the way you present yourself.
[07:38] <steveklabnik> and continuing to press forward. even when the person who wrote the damn thing says you're wrong.
[07:38] <mikekkkk> and presumably I'm presenting myself badly becuase I disaagree with you and ask you questions that challenge your view point>?
[07:38] <steveklabnik> not at all. it's the _way_.
[07:38] <mikekkkk> well you tell me a way to disagree wiht you that doesn't involve challenging you
[07:38] <mikekkkk> and I will do it that way
[07:38] <mikekkkk> and btw I don't count sugar-coating every point as a sensible solution
[07:38] <steveklabnik> dont say things like 'lols you have ocd, go get some help.'
[07:38] <mikekkkk> sometimes people need to just grow a pair and focus on the point
[07:38] <steveklabnik> even if i know you're not serious
[07:39] <steveklabnik> it takes away from an actual discussion
[07:39] <steveklabnik> it _distracts_ from a point.
[07:39] <mikekkkk> no shit I immediately clarified it was a joke because I iknew you would say that :)
[07:39] <steveklabnik> also, your discussion about if the spec is wrong or right is distracting from the original discussion, which is 'what is the spec?'
[07:39] <mikekkkk> well I think the fact that it's been revised
[07:39] <mikekkkk> pretty much puts to bed any idea that it is 'completely clear'
[07:40] <steveklabnik> sure.
[07:40] <mikekkkk> wlel then.
[07:40] <steveklabnik> that's a much better argument than 'well what about this detail and that detail and the other detail'
[07:40] <mikekkkk> I've already put that point forward several times :S
[07:40] <mikekkkk> do I need to do it in every post?
[07:41] <steveklabnik> maybe i missed it, but the first time i remember reading that is right here. that may be my fault
[07:41] <mikekkkk> yeah you did..
[07:41] <mikekkkk> I even posted it this morning in resposne to roy
[07:41] <steveklabnik> i just remember lots and lots of arguments over if an idempotent partial put is possible
[07:41] <mikekkkk> which it is
[07:41] <steveklabnik> which is irrelevant.
[07:41] <mikekkkk> what?
[07:42] <mikekkkk> it's not irrelevant it was Jan's primary fking argument
[07:42] <mikekkkk> go look at that SO thread
[07:42] <steveklabnik> I am not jan. you'll also note that i stayed out of that whole thing.
[07:42] <steveklabnik> (I think)
[07:42] <mikekkkk> well if you want to know why we had lots and lots of those arguments
[07:42] <mikekkkk> there you go
[07:42] <mikekkkk> it's nothing to do with me
[07:42] <steveklabnik> :)
[07:43] <mikekkkk> I was clarifying for the sake of people reading that that argument is based on utter fallacy
[07:43] <mikekkkk> which it is
[07:43] <steveklabnik> i dont think anyone is blameless
[07:43] <steveklabnik> i dont think you're the 'bad guy' in the discussion.
[07:43] <mikekkkk> I don't think there needs to be any blame
[07:43] <steveklabnik> exactly.
[07:43] <mikekkkk> apart from some todger called Eric
[07:43] <steveklabnik> just clarifying.
[07:43] <mikekkkk> who insists on acting like a prick and calling me a troll
[07:43] <mikekkkk> which is retarded and unnecessary
[07:43] <steveklabnik> agreed.
[07:43] <mikekkkk> ...
[07:43] <mikekkkk> cool.
[07:43] <mikekkkk> :)
[07:43] <steveklabnik> :)
[07:44] <steveklabnik> HUGS FOR EVERYONE
[07:44] <steveklabnik> lol.
[07:44] <mikekkkk> \o/
[07:46] <steveklabnik> https://github.com/rails/rails/pull/505#issuecomment-3204728 and there you go. about that.
[07:46] <steveklabnik> anyway, now i am going to go.
[07:46] <steveklabnik> because i am trying to work, in theory... haha.
[07:46] <steveklabnik> but hashing things out in real-time is good. :)
[07:46] <steveklabnik> i will see you all around from time to time.
[07:46] steveklabnik (~steveklab@c-67-165-87-244.hsd1.pa.comcast.net) left irc: Quit: bye
[07:47] <mikekkkk> hmm, that's not actually what I meant but thanks anyway
[07:48] <mikekkkk> I want to be a stateman, I don't want to be a politician.
[07:48] <mikekkkk> statesman^
[08:30] kennethreitz (~kennethre@c-24-127-96-129.hsd1.va.comcast.net) joined #rest.
[08:43] <whartung> o/
[08:50] <mikekkkk> herroooo
[08:51] <whartung> guess we had a chatty weekend
[08:56] nuclearsandwich (~nuclearsa@173.247.193.198) joined #rest.
[09:01] pc1oad1etter (~pc1oad1et@ec2-174-129-99-58.compute-1.amazonaws.com) joined #rest.
[09:06] <mikekkkk> well, mine was pretty low key
[09:06] <mikekkkk> -_-
[09:08] <whartung> we made cookies
[09:21] <darrelmiller> I packed door handles for locomotives and polished the lenses for indicator lights for locomotives. I like my income to be diversified.
[09:21] <darrelmiller> Just in case the bottom falls out of this web thing.
[09:23] impl_ -> impl
[09:29] KevBurnsJr (~KevBurnsJ@50.0.103.39) joined #rest.
[09:36] <mikekkkk> ha
[09:36] <mikekkkk> good thinking.
[09:36] <mikekkkk> can't really go wrong with locomotives too right?
[10:06] KeeperOfTheSoul -> chilversc
[10:11] mikekkkk -> mikekelly
[10:13] <darrelmiller> mikekelly: Apparently they keep building and rebuilding locomotives and probably will until we figure out how to transfer physical things over HTTP.
[10:16] <mikekelly> only a matter of time
[10:16] <darrelmiller> wonder what the media type would be called.
[10:17] <mikekelly> hal+xml
[10:17] <mikekelly> obviouslly.
[10:17] <darrelmiller> especically considering atom is taken.
[10:17] <mikekelly> i'm offended you even had to ask.
[10:17] <mikekelly> infact, I'm so offended I'm logging off.
[10:17] <darrelmiller> application/hal+particle
[10:18] <mikekelly> good bye.
[10:18] <darrelmiller> :-P
[10:18] <mikekelly> I'm gone you can't poke your tongue out at me
[10:18] <mikekelly> that;s just weird.
[10:46] mr_yall -> mryall`
[10:46] mryall` (~mryall@scifo.mexico86.co.uk) left #rest ("Killed buffer").
[11:23] <whartung> it was funny, but locally, at Knotts Berry Farm, they bought, long ago, a narrow gauge railroad out of Colorado. I got to tour their engine shops once. Basically, they just machine anything they need for the locomotives.
[11:42] bigbluehat (~bigblueha@adsl-98-71-133-72.gsp.bellsouth.net) joined #rest.
[12:16] <mikekelly> darrelmiller: seen that post to rest-discuss from jorn ?
[12:16] <mikekelly> he wants a generic media type with .NET support
[12:17] <darrelmiller> nope... I try to stay away from rest-discuss. It impacts my billable hours severly. I'll take quick look though.
[12:17] <mikekelly> http://tech.dir.groups.yahoo.com/group/rest-discuss/message/18105
[12:20] <mikekelly> I thought I'd leave it to you to reply to
[12:20] <mikekelly> since that's kind of your bag
[12:21] <mikekelly> mamund: any sage words of advice?
[12:21] <mikekelly> :P
[12:23] <mikekelly> mamund: did you see that exchange with Roy? I'm interested to know what you think
[12:23] <mikekelly> :)
[12:24] <mamund> yo
[12:24] <mamund> ??
[12:24] <mamund> Jorn is fishing for free consulting, i think<g>
[12:25] <mamund> mikekelly: Roy is playing w/ you
[12:32] <Jarda> darrelmiller: hey, mind if I ask a .NET related question?
[12:33] <darrelmiller> Jarda: Shoot.
[12:33] <Jarda> darrelmiller: I'm trying to spin my head around Async Tasks.
[12:33] <darrelmiller> Fun!
[12:33] <Jarda> So, downloading a resource from backend is a task, right?
[12:34] <darrelmiller> ... well, you could use a Task object to download a resource from a backend. Is that what you mean?
[12:34] <Jarda> so I could do var service = new MyBackend(); var task = service.FetchCustomer(123);
[12:34] <darrelmiller> The latest version of HttpClient forces you to use Tasks for every request.
[12:34] <Jarda> and then task.ContinueWith(task => { });
[12:35] <Jarda> ... task.ContinueWith(product => { });
[12:35] <Jarda> s/product/customer/ (not my tuping day)
[12:35] <darrelmiller> actually, the first one was right. task is the parameter passed to ContinueWith. You need to access task.Result to get the product.
[12:36] <Jarda> oh, true, actually
[12:36] <Jarda> so what about "the request failed"
[12:36] <Jarda> is it in task.Result also?
[12:37] <darrelmiller> Well, you mean HTTP returned a 400/500 or your client threw an exception when processing the response?
[12:37] <Jarda> kinda both
[12:38] <darrelmiller> If your client threw an exception then you will get an AggregateException being thrown somewhere. I forget exactly where you have to catch that.
[12:38] <darrelmiller> I'm still learning how to use tasks myself.
[12:38] <Jarda> I had my own threaded version with signatures like service.FetchSomeResource(Action<IResource> successCallback, Action<IResource> failureCallback);
[12:38] <darrelmiller> However, for the 400/500 then you will need to decide what to do in your FetchCustomer service method.
[12:39] <darrelmiller> HttpClient does not throw errors on 400 unlike HttpWebRequest.
[12:39] <mikekelly> mamund: by playing with me
[12:39] <Jarda> which used WebRequests internall
[12:39] <mikekelly> what do you actually mean ?
[12:39] <mikekelly> mocking me ?
[12:39] <darrelmiller> yeah, but they much catch it and swallow it.
[12:39] <darrelmiller> s/much/must
[12:40] <Jarda> darrelmiller: yeah, I meant my own implementations used WebRequests
[12:40] <Jarda> try { } catch { }
[12:41] <darrelmiller> ahh, ok, so then you need to handle those 400 exceptions and decide what it means to the FetchCustomer method. Do you rethrow or do you return a null customer object.
[12:43] <Jarda> yeah, sounds about right
[12:43] <Jarda> and some times I really wished .NET 4.5 would be out already
[12:44] <Jarda> waiting for await ..
[12:44] <darrelmiller> jarda: Here is a interesting question for you. When you use service.FetchSomeResource, are you doing a LE or LO (as per mamund's h-factors)? If you say "that depends" then can I suggest that you try and make the distinction in your client code.
[12:44] <darrelmiller> var resource = service.FetchSomeResource() works fine for LE, but in my opinion is does not match too well with LO.
[12:47] <Jarda> the h-factors were new to me, just read about them now
[12:48] <darrelmiller> h-factors are a massive piece of the puzzle when trying to grok hypermedia. Well worth the time to read up on.
[12:48] <Jarda> yeah
[12:48] <Jarda> I really need to get mamunds book
[12:49] <darrelmiller> mamund: Fantastic foreward by Leonard Richardson in your book.
[12:49] <darrelmiller> Well deserved too.
[12:49] <mamund> :)
[12:49] <mamund> i was floored, actually
[12:49] <mamund> he's working on a revision of his book, btw
[12:49] <darrelmiller> Cool.
[12:50] <mamund> Jarda: i think i did some blog posts on h-factors, too
[12:51] <Jarda> my clients are like huge sessions, which then do stuff with linked resources
[12:51] <mamund> ahh, not much help there
[12:51] <Jarda> but I don't really have any state
[12:51] <mamund> (my blog, i mean)
[12:51] <Jarda> there is just state transitions
[12:51] <Jarda> state transitions per resource
[12:51] <Jarda> I meant to say..
[12:52] <darrelmiller> jarda: What does a state transition look like in your client code?
[12:54] <Jarda> darrelmiller: hmm.. for instance I have bount my controls like <TextBox Text="{Binding Customers.Current.Name}" />
[12:55] <Jarda> and then I change the Customers.Current to another CustomerViewModel
[12:55] <Jarda> so it "transits the state"
[12:55] <Jarda> also it might be, that customer is "Locked" as in no work is to be done before he/she pays the bills
[12:56] <Jarda> so I disable controls etc if there is hypermedia missing (or pre-known "enumeration" values)
[12:56] <Jarda> but mostly my clients do know a lot about the problem domain
[12:56] <Jarda> and I'm using hypermedia to just relax my URI-design in my service
[12:57] <Jarda> so my clients know how the service works and how the problem domain behaves
[12:57] <darrelmiller> jarda: Ok, I get you.
[12:57] <Jarda> but mind you, I'm new writing hypermedia-aware clients
[12:58] <Jarda> and the actual hypermedia-aware client -problem domain is kinda new and not completely explored, as I see it
[12:59] <darrelmiller> jarda: I agree completely. There are very few examples out there of how to do it.
[12:59] <Jarda> now unfortunately I have to leave. Nice to talk to you guys once again!
[13:00] <darrelmiller> later.
[13:06] darrelmiller (~darrelmil@bas1-montreal08-1279564619.dsl.bell.ca) left #rest.
[13:24] <whartung> Holy crap -- did Eric find the Rails/PATCH thread? I just noticed it exploded to 85 messgaes.
[13:27] <mamund> :)
[13:27] <mamund> you can lay that one at mikekelly's doorstep this time<g>
[13:27] <whartung> figgers
[13:28] <mamund> LOL
[13:34] grove (~grove@84.115.45.31.customer.cdi.no) left irc: Quit: grove
[13:39] bradley-holt (~bradley-h@65-183-135-35-dhcp.burlingtontelecom.net) left irc: Quit: bradley-holt
[13:51] Wombert (~Wombert@dslb-092-074-120-056.pools.arcor-ip.net) left irc: Quit: Wombert
[14:14] bigbluehat (~bigblueha@adsl-98-71-133-72.gsp.bellsouth.net) left irc: Quit: Leaving.
[16:07] mamund sez 'bye'
[16:07] mamund (~mamund@rrcs-24-172-200-42.central.biz.rr.com) left irc: Quit: ChatZilla 0.9.87 [Firefox 8.0.1/20111120135848]
[16:30] Wombert (~Wombert@dslb-092-074-121-251.pools.arcor-ip.net) joined #rest.
[17:51] nuclearsandwich (~nuclearsa@173.247.193.198) left irc: Remote host closed the connection
[17:56] quest88 (~quest88@adsl-75-23-35-7.dsl.lgvwtx.sbcglobal.net) joined #rest.
[17:59] KevBurnsJr (~KevBurnsJ@50.0.103.39) left irc:
[18:45] quest88 (~quest88@adsl-75-23-35-7.dsl.lgvwtx.sbcglobal.net) left irc: Quit: quest88
[18:46] KevBurnsJr (~KevBurnsJ@50.0.103.39) joined #rest.
[19:32] quest88 (~quest88@75.23.35.7) joined #rest.
[21:20] Wombert (~Wombert@dslb-092-074-121-251.pools.arcor-ip.net) left irc: Quit: Wombert
[21:20] quest88 (~quest88@75.23.35.7) left irc: Quit: quest88
[22:07] sgronblo (~sgronblom@218.223.19.176) joined #rest.
[22:07] <sgronblo> I'm feeling a bit rusty about my rest
[22:08] <sgronblo> If you for example DELETE /groups/xx/users/yy is that expected to just remove that user from the group or delete the user resource completely from the system?
[22:11] grove (~grove@aggw006.cappelendamm.no) joined #rest.
[22:13] <sgronblo> Well what I feel confused about is how you are supposed to deal with associations in rest
[00:00] --- Tue Dec 20 2011