- Wednesday, August 21, 2013

json@jabber.ietf.org

Wednesday, August 21, 2013< ^ >

stpeter has set the subject to: JSON WG | http://datatracker.ietf.org/wg/json/

Room Configuration

Room Occupants

GMT+0
[13:18:04] Bjoern joins the room
[14:15:59] resnick joins the room
[14:18:59] Bjoern leaves the room
[14:37:07] m&m joins the room
[14:38:05] m&m has set the subject to: JSON WG — interim meeting 2013-08-21 < http://www.ietf.org/proceedings/interim/2013/08/21/json/proceedings.html >
[14:46:18] m&m is now known as Matt Miller
[14:48:57] cabo joins the room
[14:51:10] Paul Hoffman joins the room
[14:56:16] <Paul Hoffman> We're still early.
[14:57:51] jimsch1 joins the room
[14:58:09] <resnick> Having trouble getting into WebEx Jim?
[14:58:34] <jimsch1> My headset is not working - just rebooted
[14:59:47] <cabo> Uh, they broke it again.  "Cisco WebEx Add-On Required"
[15:00:11] <resnick> It depends on what version is currently installed on your system apparently.
[15:00:25] <cabo> Well, I can install that as often as I want.
[15:00:33] <cabo> No change.
[15:00:39] <cabo> Trying opera now.
[15:01:44] john.levine joins the room
[15:02:10] <resnick> We're waiting to begin to give people extra time to get WebEx installed.
[15:02:42] Barry Leiba joins the room
[15:04:29] <cabo> You tried to access the address https://ietf.webex.com/cmp0307l/webcomponents/widget/detect.do, which is currently unavailable.
[15:04:34] <cabo> Trying chrome now.
[15:04:58] <john.levine> works with Chrome, although on my Ubuntu box it doesn't do audio
[15:05:14] <Paul Hoffman> Those without audio can dial in.
[15:06:05] john.levine leaves the room
[15:07:03] <resnick> 1. Introduction
  What we hope to achieve
  Practicalities
2. JSON character model
3. Duplicate keys in objects
4. Next steps
[15:11:18] <resnick> Call-in User_8 == John Cowan
Call-in User_5 == ???
sdrees == Stephan ???
[15:11:54] john.levine joins the room
[15:12:07] <Paul Hoffman> sdrees = Stefan Drees
[15:12:09] <resnick> Mention the jabber room in audio.
[15:12:41] <jimsch1> Paul,  who is taking notes and are you recording?
[15:13:08] <Matt Miller> I am talking notes, and thanks for reminding about the recording!
[15:13:12] <Matt Miller> taking notes
[15:13:14] hildjj joins the room
[15:17:59] Larry Masinter joins the room
[15:18:32] <Larry Masinter> any chance of getting anyone from ECMA / TC39 to join the meeting?
[15:18:57] johnwcowan joins the room
[15:18:58] <Paul Hoffman> Larry: we announced it a few times.
[15:19:30] <Paul Hoffman> We even encouraged.
[15:19:37] <resnick> And some leadership spoke to TC-39 folks just the other day mentioning it.
[15:19:45] <resnick> Done what could be done.
[15:19:53] <Larry Masinter> so they refused to come?
[15:19:58] <resnick> (i.e., TC-39 leadership)
[15:20:02] <Paul Hoffman> Larry: we can't speak for them.
[15:20:19] <Larry Masinter> well, Crockford, will Crock be here?
[15:20:32] <Paul Hoffman> He is not on the call at this time. That's all we know.
[15:21:44] <Paul Hoffman> Larry: are you on the call as well?
[15:22:39] <Larry Masinter> no, it doesn't start for 90 minutes?
[15:22:45] <Larry Masinter> i might have messed up
[15:22:49] <Matt Miller> it's already started
[15:23:03] <Larry Masinter> sheesh sorry, i didn't do the CET calculation
[15:23:11] Stefan Drees joins the room
[15:23:35] <Larry Masinter> excuse me, i thought i had 90 minutes and i cleverly was getting jabber & sound set up in advance... i'll shut up
[15:23:44] <Larry Masinter> apologies
[15:24:42] <Paul Hoffman> Larry: to be clear, the meeting started 24 minutes ago.
[15:25:04] Matthew Morley joins the room
[15:26:06] <resnick> :) Who is wickedlogic?
[15:26:12] <Matthew Morley> Matt Morley
[15:26:18] <resnick> Hi Matt.
[15:26:25] <Matthew Morley> Hi :)
[15:29:23] <resnick> Sounds like this is wordsmithing at this point.
[15:29:36] Eliot Lear joins the room
[15:30:25] <Larry Masinter> did the idea of being explicit about layers of interpretation help?
[15:31:48] eliot.lear joins the room
[15:31:53] eliot.lear leaves the room
[15:32:06] <Larry Masinter> i'm willing to help draft about the layers, since i needed to write the stuff for IRI comparison
[15:32:23] <Larry Masinter> i struggled with the wording on IRI comparison
[15:33:10] Stefan Drees leaves the room
[15:33:39] <Larry Masinter> the two can be merged, i think
[15:35:35] <Larry Masinter> http://tools.ietf.org/html/draft-ietf-iri-comparison-02
[15:36:04] <Larry Masinter> are member names ASCII only?
[15:36:18] <Paul Hoffman> Member names are not ASCII only
[15:36:29] <cabo> No, member names are UTF-8
[15:36:30] <Larry Masinter> so they're escaped too
[15:37:46] <resnick> Speaking just as random participant: Larry, I'd like to see some straw-man proposed text for the layer stuff. I definitely hear what Joe and John are saying regarding, "People might get lost", but maybe that just means the layering discussion might make a good explanatory appendix and we stick to the ad-hoc explanation in the text itself.
[15:38:01] vinnyapp joins the room
[15:38:30] <Matthew Morley> if we are only noting the related inter-op issues, it is reasonable to state that at any layer in the model where duplicate names could come into existence, can result in inter-op issues.
[15:38:49] <Larry Masinter> yes, Pete, I'm willing to draft something based just on the general willingness. I'd like someone to look at the IRI comparison draft http://tools.ietf.org/html/draft-ietf-iri-comparison-02 and see if that makes sense
[15:39:06] <Larry Masinter> i don't like the IRI text that much
[15:39:56] <cabo> Is the intention to leave the mapping from the JSON-encoded values to strings, numbers, etc. to the applications?
[15:40:41] <jimsch1> If we are saying - act as if - then I am not sure that the model version of things workds
[15:41:07] <Larry Masinter> character normalization is a big rathole
[15:41:34] <Paul Hoffman> John only did it as analogy. A scary one.
[15:41:35] <Larry Masinter> i think it's useful to say that unicode normalization MUST NOT be done
[15:41:49] <Paul Hoffman> Lots of things MUST NOT be done.
[15:41:53] <Larry Masinter> except at the applicaiton layer
[15:42:11] <Larry Masinter> application / abstract-model / character sequence / utf8
[15:42:32] <Larry Masinter> "valid" ?
[15:43:16] <Larry Masinter> i'm also working on multipart/form-data which has the same set of issues, since it's an attribute/value pair collection
[15:43:25] <cabo> Right out of a JSON-using spec:
[15:43:26] <cabo>     1.    JSON native types, true, false, nil, String, Hash, Array, and Number are encoded normally.
    2.    A Symbol is encoded as a JSON string with a preceeding `:` character.
    3.    The `^` character denotes a special key value when in a JSON Object sequence.
    4.    A Ruby String that starts with `:` or the sequence `^i` or `^r` are encoded by excaping the first character so that it appears as `\u005e` or `\u003a` instead of `:` or `^`.
[15:43:50] <cabo> I would argue that is not an interoperable usage of JSON.
[15:44:37] <Larry Masinter> if that's wrong, then it isn't an instance of application/json
[15:44:59] <resnick> I'm waiting for the word "semantic" to appear in this discussion.
[15:45:00] <Larry Masinter> cabo, i agree
[15:45:17] <Larry Masinter> resnick, we're using 'application' instead, which is better
[15:45:55] <Paul Hoffman> Ah, hardware. I'm now on audio.
[15:46:39] <resnick> I think that's what we're talking about: The semantic/meaning of "\0065" is identical to the semantic/meaning of "A"
[15:46:52] <cabo> ("e", actually :-)
[15:47:03] <Larry Masinter> resnick, people can imbue semantics to anything, as with cabo's example
[15:47:07] <resnick> Damn. Hex/decimal!
[15:47:55] <Larry Masinter> i'm not sure what i should draft, if someone wanted me to write about layers ?
[15:48:41] <Larry Masinter> did anyone talk about normatively referencing the ECMA JSON document?
[15:49:01] <cabo> What document is that?
[15:49:14] <Larry Masinter> the one that i was hoping someone from TC39 would come talk about
[15:49:25] <cabo> Oh.  URI?
[15:49:27] <Larry Masinter> if anyone had actually talked to anyone
[15:49:30] <Larry Masinter> i don't have a URI
[15:52:16] <Larry Masinter> seems twisted, but at least you talked to them
[15:53:18] <Larry Masinter> i'll take this offline
[15:53:56] <resnick> Happy to discuss it with you Larry.
[15:54:10] <johnwcowan> Larry, I think it would amount to a fairly broad redrafting to do so.
[15:55:58] <Larry Masinter> johnwcowan, "so" is ambiguous. I'm thinking of a few sentences not more than 8 sentences, and then a review of the rest of the document to adjust wording
[15:57:28] <Larry Masinter> "You can think of JSON processing as operating on several levels: an encoding (UTF-8), a sequence of Unicode characters, an absstract data model, and an application model. When talking about comparison, it is meaningful to describe it at the abstract data model. A JSON processor ...   "    something like that as the start
[16:00:14] <Larry Masinter> does the abstract model allow it?
[16:05:15] <Paul Hoffman> Off-topic: Vinny, can you say what your last name is for the minutes?
[16:05:27] <cabo> You can't write those numbers in decimal.
[16:06:02] <Larry Masinter> the problem is that people have used the JSON framework and syntax and mapped it onto a different abstract model, and they don't understand why the working group doesn't allow it (like duplicates).
[16:06:15] <Larry Masinter> s/working group/spec/
[16:07:44] <Larry Masinter> https://github.com/datalanche/json-bignum
[16:08:18] <Larry Masinter> i think number precision and representation has to be at the application layer
[16:09:06] <cabo> They might be handling 20-byte strings only.
[16:09:22] <cabo> That is kind of out of scope.
[16:09:51] <Larry Masinter> i take back what i just said, sorry, reading more
[16:09:53] <cabo> Pete is right about the difference between quantities and broken usage.
[16:11:52] <jimsch1> What about implementations which handle integers and floats differently for percision?
[16:11:57] <vinnyapp> (Off-topic) Sorry for the delay Paul, was focusing on the phone chat. My last name is "An". I'm slot 5 on the call
[16:13:01] <Paul Hoffman> Thanks, Vinny.
[16:14:01] <cabo> Yep, needs to be fixed.
[16:14:43] <Larry Masinter> i think this is application problem
[16:14:52] <Larry Masinter> implementations may set limits on lots of things
[16:16:11] <Larry Masinter> implementations are likely to set limits on numbers
[16:16:27] <Larry Masinter> RFC4627 doesn't distinguish between integers and floating point
[16:16:43] <Larry Masinter> "Implementations are LIKELY to set limits"
[16:17:19] <Larry Masinter> it shouldn't be a normative reference
[16:17:47] <resnick> Agree to both what Paul said and what Larry said.
[16:18:06] <resnick> This is implementation advice that appears in section 4.
[16:18:10] <resnick> Not normative stuff.
[16:19:33] <Larry Masinter> it's bike-shedding
[16:19:51] vinnyapp leaves the room
[16:20:27] <Larry Masinter> i'm off for dinner too now
[16:20:36] <Larry Masinter> i'm not a european but i'm playing one this week
[16:23:09] <cabo> https://github.com/rwaldron/tc39-notes/blob/master/es6/2013-07/Draft%20Standard%20The%20JSON%20Data%20Interchange%20Format.pdf
[16:23:18] <resnick> John Cowan, please mute.
[16:25:29] tonyhansen joins the room
[16:28:29] vinnyapp joins the room
[16:32:45] <Larry Masinter> i'm not interested in the second two
[16:36:18] <Larry Masinter> i'm in a small town in Hungary, good internet
[16:36:28] <Larry Masinter> "Tim Bray" profile :)
[16:37:42] <Larry Masinter> "Because it is so simple, it not expected that the JSON grammar will ever change."
[16:39:04] <Larry Masinter> the document Cabo posted is a good read, and it's public
[16:40:05] <Larry Masinter> want to talk about the BCP 70 update
[16:44:24] <Larry Masinter> it's a honey pot
[16:44:32] <resnick> John Levine, please mute.
[16:44:48] <Larry Masinter> i better run off now anyway
[16:44:56] <Matthew Morley> as a side note, php's json_decode only supports utf8 according to the docs page
[16:44:58] <john.levine> sorry, thought I was muted
[16:45:12] <Larry Masinter> ok, see you
[16:45:34] Larry Masinter leaves the room
[16:48:12] hildjj leaves the room
[16:48:15] Barry Leiba leaves the room
[16:48:20] john.levine leaves the room
[16:48:31] vinnyapp leaves the room
[16:48:34] Matt Miller leaves the room: Disconnected: connection closed
[16:49:05] Matthew Morley leaves the room
[16:49:11] <Paul Hoffman> Thanks again, all.
[16:49:41] Paul Hoffman leaves the room
[16:49:55] resnick leaves the room
[16:51:33] m&m joins the room
[16:54:48] Eliot Lear leaves the room: Replaced by new connection
[16:54:52] Eliot Lear joins the room
[16:56:12] hildjj joins the room
[17:03:14] jimsch1 joins the room
[17:05:03] johnwcowan leaves the room
[17:05:57] hildjj leaves the room
[17:05:57] cabo leaves the room
[17:05:57] jimsch1 leaves the room
[17:05:57] jimsch1 leaves the room
[17:17:12] Eliot Lear leaves the room
[18:18:43] Tim Bray joins the room
[18:19:06] Tim Bray leaves the room
[21:03:23] tonyhansen leaves the room
[22:01:07] m&m leaves the room: Disconnected: connection closed