IRC log of bpwg on 2007-11-06

Timestamps are in UTC.

13:32:12 [RRSAgent]
RRSAgent has joined #bpwg
13:32:12 [RRSAgent]
logging to
13:35:48 [jo]
Meeting: Mobile Web Best Practices Working Group F2F Day 2
13:35:57 [jo]
Date: 6 Nov 2007
13:36:03 [jo]
Chair: Dan, Jo
13:58:43 [francois]
francois has joined #bpwg
14:01:05 [edm]
edm has joined #bpwg
14:01:23 [PhilA]
PhilA has joined #bpwg
14:01:49 [PhilA]
DKA: Begins meeting, suggests tour de table...
14:02:34 [DKA]
DKA has joined #bpwg
14:04:13 [jo]
Present: Drew_Major, Dave_Raggett, Rhys_Lewis, Francois, SeanP, MikeSmith, Kai, PhilA, Bryan, Aaron_Kemp, Jose, Ileana, Serge, Lymm, EdM
14:04:47 [PhilA]
Dan and Jo sort out the logistics, scribes, agendas and so on
14:05:25 [SeanP]
SeanP has joined #bpwg
14:05:46 [MikeSmith]
MikeSmith has joined #bpwg
14:06:35 [Rhys]
Rhys has joined #bpwg
14:06:48 [edm2z]
edm2z has joined #bpwg
14:07:09 [PhilA]
agenda+ Content Trnsformation
14:07:35 [PhilA]
agenda+ Issues and actions
14:07:39 [PhilA]
agenda+ mobileOK
14:08:25 [marie]
marie has joined #bpwg
14:08:44 [PhilA]
DKA: How many people here in CT TF?
14:08:51 [PhilA]
four hands
14:09:14 [PhilA]
ack four
14:09:23 [soonho]
soonho has joined #bpwg
14:09:29 [jo]
Present+ Mark_Baker, Sean_Own, Soon-Ho_Lee
14:09:37 [PhilA]
scribeNick: PhilA
14:09:42 [PhilA]
scribe: Phil
14:09:52 [PhilA]
There are 4 CT TF members present
14:10:35 [edm2z]
14:10:45 [MikeSmith]
14:10:50 [IleanaLeuca]
IleanaLeuca has joined #bpwg
14:11:00 [achuter]
achuter has joined #bpwg
14:11:03 [PhilA]
Rhys:CT TF has 2 deliverables as decided in London f2f
14:11:03 [MikeSmith]
RRSAgent, please draft minutes
14:11:03 [RRSAgent]
I have made the request to generate MikeSmith
14:11:17 [PhilA]
Rhys: 1 - to define problem statemetn (limited scope)
14:11:35 [PhilA]
Rhys: guidelines document to help people avoid the pitfalls
14:11:47 [PhilA]
14:11:57 [DKA]
14:12:03 [PhilA]
Rhys: Problem doc is now in TR space and will become a WG Note
14:12:16 [edm2z]
14:12:30 [PhilA]
Rhys: Where there are multiple players in delivery channel between server and client
14:12:51 [PhilA]
Rhys: Is trying to work out how they can signal to each other as they don't cuase errors in what each other does
14:13:06 [PhilA]
Rhys: eg - material on origin server is already adapted for the end device
14:13:24 [PhilA]
Rhys: Found recent situations where intervening transforming proxy has carried out inappropriate transformations
14:13:38 [Kai]
Kai has joined #bpwg
14:13:42 [PhilA]
Rhys: purpose of TF is to work out what can flow between participants to avoid that happening
14:13:52 [jo]
Present+ Rob_Finean, Jonathan_Jeong
14:13:55 [PhilA]
Rhys: Focussed on using existing W3C tech to do it
14:14:12 [MikeSmith]
14:14:17 [PhilA]
Rhys: Therefore we're producing a guidelines doc, not a normative Rec
14:14:29 [PhilA]
Bryan: Do no harm, not do it better?
14:14:31 [PhilA]
Rhys: Yes
14:14:37 [srowen]
srowen has joined #bpwg
14:15:01 [MikeSmith]
14:15:02 [PhilA]
Rhys: Who's read the doc?
14:15:05 [MikeSmith]
RRSAgent, please draft minutes
14:15:05 [RRSAgent]
I have made the request to generate MikeSmith
14:15:07 [PhilA]
Many hands go up
14:15:24 [PhilA]
Rhys: Notes Bryan's comments
14:15:43 [MikeSmith]
14:15:50 [PhilA]
Rhys: we have a skeleton guidelines doc and we're working on filling in the blanks
14:16:06 [PhilA]
Rhys: Jo has recently done some work on suitable signalling
14:16:30 [PhilA]
Rhys: We're not just talking about proxies and servers - clients have a role to play in user preferences
14:17:01 [MikeSmith]
14:17:03 [PhilA]
Rhys: User should have choice in whether they see transformed or raw content
14:17:25 [PhilA]
Rhys: Invites Jo to take us through his recent post
14:18:08 [edm2z]
14:18:21 [jo]
-> Action Relating to Possible Techniques
14:18:23 [PhilA]
DKA: A question in the meantime - a lot of this was kicked off by the controversy about content adaptation proxies not sending the UA string through
14:18:58 [PhilA]
DKA: Are we going to give content providers the guidance they need or will it be so elaborate that it won't help?
14:19:15 [PhilA]
Rhys: It wouldn't be productive if we couldn't find a relatively simple solution
14:19:17 [chaals]
chaals has joined #bpwg
14:19:37 [jo]
Present+ Abel
14:19:39 [IleanaLeuca]
IleanaLeuca has joined #bpwg
14:19:47 [PhilA]
Rhys: The feeling I get is that there is a fighting chance of coming up with some helpful guidance
14:19:56 [jo]
Present+ Arun
14:20:09 [PhilA]
DKA: My litmus test - I'd like to be able to talk to developers without being booed down as happens now...
14:20:30 [Jonathan]
Jonathan has joined #bpwg
14:20:35 [PhilA]
DKA: More seriously, you need not worry - here is what we, the industry, are doing to fix this problem
14:20:35 [jcantera]
jcantera has joined #bpwg
14:21:14 [PhilA]
Rhys: The idea that we'll provide something that can please most of the people most of the time is our aim
14:21:45 [PhilA]
DKA: The message is - 'adapt your content' - that's the way forward for the mobile web?
14:21:55 [PhilA]
DKA: The best outcome from this doc is that it fits into that univese
14:22:04 [lynne]
lynne has joined #bpwg
14:22:05 [rob]
rob has joined #bpwg
14:22:10 [PhilA]
Rhys: We're articulating how transforming proxies fit into that eco system
14:22:13 [Rhys]
14:22:20 [Rodrigo]
Rodrigo has joined #bpwg
14:22:28 [PhilA]
Bryan: The message that developers get - we need to educate them to the nature of the options that they have
14:22:52 [PhilA]
Bryan: Either very specific to mobile, or adpating to mobile, or not caring and leaving it up to adaptation
14:23:11 [PhilA]
Bryan: We need to educate them what different headers can do and why they should use them - and then move on
14:23:51 [PhilA]
Jo: There are various options. One very narrow using HTTP and the other related to the scope of the guidelines doc
14:24:41 [PhilA]
Slight hiatus while Jo and Dan do a jog for us all
14:25:41 [PhilA]
Jo: This is a very dense debate. Bryan in particular, made some very useful comments...
14:25:57 [PhilA]
14:26:18 [Rhys]
14:26:20 [jo]
14:26:28 [Jonathan]
Jonathan has left #bpwg
14:26:38 [PhilA]
Jo: The question is - how far do you go?
14:27:06 [PhilA]
Jo: Are we trying to write a doc that tells people how not to do any harm, or are we going to guide people through the whole chain?
14:27:27 [PhilA]
Jo: The consensus seems to be that we're not going all the way to the detail extreme
14:27:46 [PhilA]
Jo: Complicated by some browsers that act as proxies :-)
14:28:04 [PhilA]
Jo: A sequence of proxies may be treated as a unit of proxy capability
14:28:13 [abel]
abel has joined #bpwg
14:28:38 [PhilA]
Jo: The issue at hand is - are we going to deliver the desktop presentation or the specifically mobile-crafted version and where is that latgter version created, if it is created
14:29:06 [Kai]
14:29:11 [PhilA]
Jo:Bearing in mind that the web kits browser (??) can do a lot, how does the server know what the browsser is able to do?
14:29:57 [PhilA]
Jo: Given that the browser is not able to provide the mobile experiecne on its own and the server may be able to, how do we get the mix right?
14:30:14 [PhilA]
Jo: There are 64 permutations I reckon
14:30:28 [PhilA]
Jo: Each of the 3 components can do one of 2 types of transformation
14:30:44 [Jonathan]
Jonathan has joined #bpwg
14:30:51 [PhilA]
Jo: At the markup level they can clean up markup, compress images etc. This seems different from re-formatting of content, including or exclusing some etc.
14:31:03 [PhilA]
14:31:23 [MikeSmith]
s/web kits browser (??)/WebKit [which both the Nokia S60 and Apple Safari and iPhone browsers use as their backend]/
14:31:45 [PhilA]
Rhys: There are lots of options for how the differnt entities can react to each other
14:32:06 [PhilA]
Rhys: The complete set of options is large
14:32:07 [Bryan]
Bryan has joined #bpwg
14:32:30 [PhilA]
Rhys: e.g. is the server can or can't negotiate to come up with a mobile version etc
14:33:01 [PhilA]
Rhys: we should minimise the number of things we address in the initial doc as we have very limited time for the TF
14:33:20 [PhilA]
Rhys: Then again, if we miss out key use cases, then we'll be skipping too lightly
14:33:40 [PhilA]
Bryan: I can add a summary of the conversation...
14:34:01 [PhilA]
Bryan: Let me try and get some yes/no in/out statements
14:34:08 [PhilA]
Bryan: Whetehr to offer choice to user? Yes
14:34:14 [PhilA]
Bryan: How to offer chocie? No
14:34:20 [PhilA]
Bryan: Behaving as requested? Yes
14:34:33 [PhilA]
Bryan: Protecting from harm? Yes
14:34:42 [PhilA]
Bryan: Avoid non-web app harm? Yes
14:34:53 [PhilA]
Rhys: Is this in an e-mail?
14:35:01 [PhilA]
Bryan: yes, but this is a summary
14:35:13 [arun]
arun has joined #bpwg
14:35:18 [PhilA]
Bryan: Objectives - usability and interop - yes
14:35:36 [PhilA]
Bryan: Optimisation, personalisation, content control - grey areas
14:35:51 [PhilA]
Bryan: If it becomes a policy point then I'd say No
14:36:28 [PhilA]
Jo: Good summary of teh kinds of things we've been pondering - broadly I agree
14:36:37 [PhilA]
14:36:49 [PhilA]
Jo: Might be worth working through some use cases
14:37:19 [PhilA]
Jo: Case 1 - the classic case that the server only has a deskptop only presentation and the server only has desktop
14:37:29 [achuter]
q+ to mention mobileOK complication
14:37:50 [PhilA]
Jo: In this case it is possible that the server's presentation may be universal and so will lay out properly in either
14:37:57 [PhilA]
Jo: In general that's not the case
14:38:04 [Kai]
q+ to ask about layout independent presentation
14:38:39 [PhilA]
Jo: A personal feeling - proxies should do what they do reluctantly, more in sorrow than anger.
14:38:46 [DKA]
14:38:55 [PhilA]
ack alan
14:39:01 [PhilA]
ack achut
14:39:01 [Zakim]
achuter, you wanted to mention mobileOK complication
14:39:20 [PhilA]
ack kai
14:39:20 [Zakim]
Kai, you wanted to ask about layout independent presentation
14:40:00 [PhilA]
Kai: I noticed a potential use case missing - a layout-neutral presentation. Maybe an XML instance provides the data and the layout is separate
14:40:24 [Rhys]
14:40:28 [PhilA]
Jo: Yes, but I think what we're trying to do, erm, I don't think that's a use case that we want to consider at the moment
14:40:52 [PhilA]
Jo: Stylesheet application etc. But that's not what we're talking about within our scope with limited timeline
14:41:21 [PhilA]
Jo: Case 2 - server has both mobile and desktop presentation available
14:41:58 [PhilA]
Jo: Assuming that the content provider's wishes are honoured, the user will get the mobile presentation - with the caveat that there should be a way for the user to specify the desktop presentationb
14:42:13 [SeanP]
14:42:23 [PhilA]
Jo: If both content provider and user both express preferences and they're different, who wins?
14:42:40 [PhilA]
Jo: CP may be able to say I don't care what you want, I'm sending you this
14:42:54 [PhilA]
ack: sea
14:43:17 [PhilA]
seanP: Usually the suer preference will prevail
14:43:23 [PhilA]
14:44:07 [PhilA]
seanP: There's some discussion about whetehr the mobile page is better than the transformed desktop page, but sometimes the desktop page may have more content that the user actually wants
14:44:29 [Kai]
q+ to speak of prior experiences with providing desktop content for mobile users
14:44:35 [PhilA]
SeanP: If the mobile version is a summary, there are cases where the suer wants the full content, irrespective of the fact it looks awful...
14:44:46 [PhilA]
ack: sean
14:44:55 [PhilA]
ack SeanP
14:45:25 [PhilA]
Rhys: Where we do have a user preference, it's pretty straightforward - we should honour the user's preference
14:45:52 [PhilA]
Rhys: Where the user doesn't have the ability doesn't have the choice then the author gets the next shout
14:46:14 [Rhys]
14:46:15 [PhilA]
Rhys: (i.e. if you don't know what the user wants)
14:46:26 [Rhys]
ack kai
14:46:26 [Zakim]
Kai, you wanted to speak of prior experiences with providing desktop content for mobile users
14:46:26 [jo]
ack k
14:46:31 [PhilA]
Kai: We've tried for a long time to take desktop content and make it work for mobile - and it doesn't work
14:46:48 [PhilA]
Kai: User choice would be good - and we should store that choice for next time
14:47:26 [DKA]
14:47:28 [PhilA]
Kai: On an editorial level, there's far more info on a desktop page than a typical mobile page. We've tried to create condensed versions automagically and it doesn't work - you have to create 2 separate instances
14:47:33 [Rhys]
acl dka
14:47:35 [jo]
ack dka
14:47:41 [PhilA]
DKA: We're straying into policy
14:47:59 [Rhys]
14:48:04 [PhilA]
DKA: Shouldn't we be transparent for all parties so everyone knows what's going on?
14:48:52 [jo]
14:49:02 [PhilA]
DKA: If you have enough info so the browser, proxy, server and user know what's available, then you can use business rules to work out waht to do
14:49:07 [PhilA]
14:49:10 [DKA]
14:49:28 [PhilA]
Rhys: Interesting. That speaks to the guidelines scope
14:49:34 [jo]
q+ bryan
14:50:18 [PhilA]
Rhys: It should probably say that if you don't know the user preferences you should use the adapted conte,t not transcoded contnet. Then you might expose that policy as a user-configurable option - but that sounds like a level up from where we're at
14:50:27 [PhilA]
Jo: I agree
14:50:28 [Rhys]
ack jo
14:50:33 [Rhys]
ack rhys
14:50:37 [Rhys]
14:50:46 [PhilA]
Jo: I don't thinkwe should specify policy, but we should illuminate that there are policy choices to be made
14:50:55 [PhilA]
Jo: ANd we should provide a vocabulary to enact those policies
14:51:34 [PhilA]
Jo: e.g. we might say 'in general, CP, it is bad practice to be hard-nosed about re-formatting, especially in terms of removal of bad mark-up etc.
14:52:04 [PhilA]
Jo: However, I don't think we should remove the option from the CP to say 'I know what I'm doing'.
14:52:36 [PhilA]
Jo: Even if they don't, that's probably not the point. It's their content and ultimately, that's the war
14:53:35 [PhilA]
Jo: SO we can summariese the policy choices, here's the vocab to express those policies, and we think you should use option x and leave it at that
14:54:06 [PhilA]
Jo: If the operator of the proxy thinks that untransformed content will cuase a problem then the user should be warned
14:54:23 [SeanP]
14:54:31 [Kai]
ack bry
14:54:40 [PhilA]
Bryan: As Rhys said, we're trying to solve the problems, not opening up new ones. The key thing is how to express the desire?
14:54:57 [Rhys]
ack bryan
14:55:02 [PhilA]
Bryan: There has to be some understanding of intent
14:55:15 [PhilA]
Bryan: We need to figure out how to signal intent in a dynamic way
14:55:32 [PhilA]
Rhys: Yes - we need to work out techniques for that and that should work in teh absence of intent as well
14:55:37 [PhilA]
ack sean
14:55:39 [Rhys]
ack seanp
14:56:25 [PhilA]
SeanP: The proposed 'do not transform' header does that, but I think there should be another like 'do not transform unless you know you can do a better job' or whatever
14:56:35 [Kai]
q+ to mention the conflict of user choice vs. provider choice. Both needs to be possible, but may not be easy to do simultaneously. Give providers tools to do what is necessary, but don't proscribe
14:56:37 [jo]
14:56:38 [PhilA]
Sean: There's a limited number of options but it's more than binary
14:56:44 [PhilA]
ack Kai
14:56:44 [Zakim]
Kai, you wanted to mention the conflict of user choice vs. provider choice. Both needs to be possible, but may not be easy to do simultaneously. Give providers tools to do what
14:56:49 [Zakim]
... is necessary, but don't proscribe
14:57:32 [PhilA]
Kai: Users may well not understand the choices...
14:57:48 [Rhys]
14:57:50 [PhilA]
Kai: There's a need to give CPs the tools to do it right. Whetehr they use them or not is an open question
14:58:40 [PhilA]
rrsagent, draft minutes
14:58:40 [RRSAgent]
I have made the request to generate PhilA
14:58:52 [edm2z]
scribe: edm
14:59:07 [edm2z]
scribenick: edm2z
14:59:46 [kemp]
q+ to talk about the iphone
14:59:58 [Rhys]
15:01:16 [edm2z]
jo: whose preferences should take priority is not always a straightforward question ...
15:01:51 [SeanP]
15:02:01 [kemp]
15:02:07 [Rhys]
15:02:36 [Magnus]
Magnus has joined #bpwg
15:02:38 [jo]
[note to self - bookmark the "roaming issue"]
15:02:50 [edm2z]
rhys: operator and user position may be influenced by cost issues
15:03:32 [Rhys]
ack sean
15:03:47 [Rhys]
15:03:52 [edm2z]
DKA: e.g., user may be happy to use desktop version with fast data service, mobile otherwise ...
15:04:30 [Kai]
15:04:36 [jo]
ack rh
15:05:25 [edm2z]
SeanP: mobile browsers may be unable to deal with some types of content (e.g., Flash), but proxies could adapt these...
15:05:44 [jo]
ack k
15:05:46 [Rhys]
15:06:07 [SeanP]
15:06:37 [Rhys]
ack sean
15:06:37 [edm2z]
rhys: we do not want to encourage using separate URIs for accessing desktop vs. mobile friendly content
15:06:57 [jo]
[kai sometimes neither the cp or the suer knows what hey want, are we going to provide some guidelines]
15:07:15 [edm2z]
15:08:26 [edm2z]
rhys: everyboby should now have a good idea of the anticipated scope of CT guidelines
15:09:16 [jo]
-> Possible Techniques for signalling capability/awares
15:09:21 [Bryan]
15:09:39 [jo]
s/awares/awareness or preference
15:09:41 [Rhys]
ack bryan
15:09:54 [edm2z]
PhilA: content personalization is out-of-scope, but could it be considered to be a useful extension of CT ?
15:10:40 [edm2z]
Bryan: we should narrow the primary focus of CT to interoperability and usability
15:12:02 [edm2z]
rhys: initial scope is very narrow - to solve a particluar problem - but we do not want to limit potential future functionality of adaptation proxies
15:13:03 [edm2z]
jo: let's start discussing specific techniques - see
15:15:03 [edm2z]
jo: refers to the list in
15:15:31 [Rhys]
15:16:28 [tskytta]
tskytta has joined #bpwg
15:17:22 [Rhys]
15:17:37 [edm2z]
jo: [continues to review the list above - in the spirit of HTTP headerspotting??]
15:17:49 [kemp]
15:18:26 [PhilA]
15:18:49 [Rhys]
ack me
15:19:09 [jo]
ack aa
15:19:14 [jo]
ack kemp
15:19:31 [Bryan]
15:19:37 [edm2z]
rhys: these techniques may not be universally usable - can a transforming proxy be informed by the UA header what to do ?
15:19:46 [srowen]
15:20:15 [jo]
ack phila
15:21:39 [Rhys]
15:22:05 [edm2z]
PhilA: huge amount of info may have to be crammed into a couple of headers - should we focus on what info would be most useful?
15:22:46 [edm2z]
PhilA: other W3C groups are already working on the issue of how to communicate certain info ...
15:24:01 [arun_]
arun_ has joined #bpwg
15:24:25 [jo]
15:24:27 [edm2z]
Bryan: we should come up with a simplest possible solution to the problem - not necessarily overload the existing semantics
15:24:28 [Rhys]
ack bryan
15:24:36 [jo]
ack sro
15:25:36 [edm2z]
srowen: simple is best - we should recommend not changing original UA
15:26:18 [kemp]
[actually sean is advocating _changing_ the user agent to the proxy's useragent]
15:27:43 [srowen]
(srowen paraphrases himself: proxy should not do anything but say "I am a proxy" in its User-Agent. Therefore don't recommend anything but what proxies are doing today for the request. No need to overload, stretch, or invent mechanisms. 20, 21, and 23 are existing simple way to communicate "I may adapt content". Proxy merely redirects to origin server and gets out of the way. So 20, 21, 23 seem to solve it all nicely to me.)
15:27:49 [Rhys]
ack me
15:27:56 [PhilA]
ack jo
15:28:08 [edm2z]
rhys: we try to overloading existing HTTP semantics - but educating about its proper use makes sense
15:28:34 [Rhys]
15:28:49 [Rhys]
ack me
15:29:49 [edm2z]
jo: we are not chartered to create new technology - should leverage what exists, bt also point out any constraints we may be running into
15:33:37 [MikeSmith]
Present+ KlausBirkenbihl, JonFerraiolo, Timo, Rodrigo, AnnBasetti, GeoffFreed
16:04:25 [Rodrigo]
Rodrigo has joined #bpwg
16:09:15 [Bryan]
Bryan has joined #bpwg
16:09:39 [MikeSmith]
Scribenick: MikeSmith
16:10:40 [MikeSmith]
Rhys : We just completed discussion on what role and additional User-Agent header might play.
16:10:57 [Rhys]
16:11:34 [MikeSmith]
Topic: Add the real user agent as part of the massaged user agent string
16:11:57 [MikeSmith]
kemp : I think putting it in the User-Agent is probably a bad idea ...
16:12:19 [MikeSmith]
... I'd be more inclined to change the User-Agent to say "Inspect this other header"
16:13:03 [MikeSmith]
Topic: Tasting of content with original header.
16:13:06 [abel_]
abel_ has joined #bpwg
16:13:11 [klaus]
klaus has joined #bpwg
16:13:45 [MikeSmith]
jo : question of idempotency and other questions ...
16:14:30 [Rhys]
16:15:22 [MikeSmith]
... the other advantage of this is that it addresses the concerns of the those who don't want to change their current [habits/expectations with regard to headers]
16:16:04 [MikeSmith]
Topic: Use an Expect header
16:16:19 [MikeSmith]
jo : this is a little rococo ...
16:16:48 [MikeSmith]
... certainly fits in with the existing rules of HTTP
16:18:02 [Rhys]
16:18:08 [MikeSmith]
Option 7: Embed original HTTP headers as part of the multipart MIME-encoded elaboration of the request as a message/http part
16:18:23 [MikeSmith]
RRSAgent, please draft minutes
16:18:23 [RRSAgent]
I have made the request to generate MikeSmith
16:19:11 [MikeSmith]
Rhys: Are we aware of any examples of where passing other headers through causes any problems?
16:19:17 [Rhys]
ack me
16:19:22 [MikeSmith]
jo : The are two separate cases ...
16:20:03 [SeanP]
16:20:52 [MikeSmith]
Rhys : question is which headers are actually expected and can we find a way to pass them through.
16:20:56 [DKA]
ack s
16:21:59 [MikeSmith]
Point 8. Extend the interpretation of host or nickname field in the Via header
16:22:15 [MikeSmith]
jo : HTTP says you can drop comments from the Via header ...
16:22:18 [SeanP]
16:22:40 [Rhys]
The point was that there are other headers that an origin server might use. An alternative mechanism to the multipart mime approach is simply not to replace those other headers. The decision on whether or not to repace other headers is the same one as whether or not to replace the user agent header
16:22:41 [MikeSmith]
... so information can be lost and we can't rely on it being preserved
16:22:51 [SeanP]
ack me
16:23:10 [MikeSmith]
jo : Question is, are these comment parts every actually dropped [in practice]
16:23:30 [MikeSmith]
Alternative is to use an X-Via header
16:23:49 [MikeSmith]
... which is defined as a copy of the Via header but with comments always preserved
16:24:10 [arun_]
arun_ has joined #bpwg
16:24:21 [Rhys]
16:24:47 [Bryan]
16:24:55 [MikeSmith]
so all the above was a discussion related of Request [stuff]
16:25:15 [DKA]
ack b
16:25:20 [MikeSmith]
now on discussion of Response stuff
16:25:29 [MikeSmith]
Topic: Response
16:27:05 [Rhys]
16:27:10 [Rhys]
16:27:35 [DKA]
ack r
16:28:09 [MikeSmith]
jo : As a content provider, you may be aware of potential markup cases that could cause problems with certain handsets but you don't fix those because you choose to leave it up to CT proxies or something to deal with instead.
16:28:36 [MikeSmith]
Rhys : CT proxies are providing value by solving those kinds of problems.
16:28:45 [SeanP]
16:28:58 [DKA]
ack s
16:29:03 [Rhys]
16:29:15 [MikeSmith]
Rhys : In this TF we are talking about CT proxies that transform the current Web
16:29:35 [MikeSmith]
Bryan : You can extend Cache-Control to more than just No-Transform
16:31:59 [MikeSmith]
Point 22: Taste the content
16:32:22 [MikeSmith]
Point 23: Look for link elements
16:32:30 [PhilA]
16:32:32 [Rhys]
16:33:13 [kemp]
16:33:27 [PhilA]
16:33:44 [DKA]
ack k
16:34:27 [MikeSmith]
kemp : The "look for LINK elements" practice is out there and widely used ...
16:34:36 [MikeSmith]
... so we do need to acknowledge it
16:36:20 [SeanP]
16:37:09 [DKA]
ack s
16:37:13 [MikeSmith]
SeanP : back to 300 response ...
16:37:40 [MikeSmith]
... question: Is it used already with some default body that we would end up stomping on?
16:38:54 [Bryan]
16:40:06 [Rhys]
ack bryan
16:40:42 [MikeSmith]
Bryan : Are we chartered to create new uses, new headers?
16:40:52 [MikeSmith]
Rhys : My view is no, we are not.
16:41:10 [MikeSmith]
... we might conclude that we can't do this just by producing a guidelines document ...
16:41:36 [MikeSmith]
... but there is at least a fighting chance that some combination of them will in fact do what we need ...
16:42:18 [DKA]
q+ on 300
16:42:38 [Rhys]
ack 300
16:42:52 [Rhys]
ack on
16:43:20 [MikeSmith]
DKA : If we do decide to do something with 300, we need to be very sure to [do thorough research on it]
16:43:21 [jo]
16:43:24 [jo]
q+ ph
16:43:29 [jo]
ack ph
16:43:33 [MikeSmith]
... get some findings
16:43:38 [PhilA]
16:44:36 [PhilA]
q+ to quote from HTTP WG charter: "The WG is not tasked with producing
16:44:36 [PhilA]
new methods, headers, or extension mechanisms, but may introduce new
16:44:36 [PhilA]
protocol elements if necessary as part of revising existing
16:44:36 [PhilA]
functionality which has proven to be problematic"
16:44:48 [Rhys]
ack jo
16:45:03 [MikeSmith]
jo : whatever we do come up with will require a thorough period of testing ...
16:45:04 [Rhys]
16:45:42 [PhilA]
ack me
16:45:42 [Zakim]
PhilA, you wanted to quote from HTTP WG charter: "The WG is not tasked with producing
16:47:13 [MikeSmith]
Rhys : Anybody have anything to add to the list?
16:47:25 [MikeSmith]
DKA : We are focusing on the server and network side ...
16:47:31 [kemp]
16:47:35 [jo]
16:47:36 [Rhys]
16:47:46 [MikeSmith]
... but so far, not address the case of advanced browsers ...
16:48:02 [MikeSmith]
... being able to communicate [their advanced-ness]
16:48:37 [MikeSmith]
kemp : I think some of these do address the case of allowing a browser to communicate its capabilities
16:48:54 [MikeSmith]
jo : I think there are some things outside of what Aaron already mentioned ...
16:49:12 [MikeSmith]
... potential use or misuse of Pragma, for example
16:49:37 [MikeSmith]
Rhys : A mechanism for transmitting the notion that the user has made a choice ...
16:49:43 [Rhys]
16:49:44 [MikeSmith]
... about what he or she wants to see
16:49:47 [kemp]
16:49:55 [jo]
ack me
16:49:56 [Rhys]
ack jo
16:49:59 [Rhys]
ack me
16:50:13 [MikeSmith]
DKA : What I was talking about was more on the guidelines side of things
16:51:44 [MikeSmith]
DKA : Also, I'm worried about this "tasting" thing ...
16:51:56 [jo]
16:52:10 [Bryan]
16:52:16 [MikeSmith]
[DKA mentions problem of someone putting an item into shopping cart twice]
16:52:41 [PhilA]
q- jo
16:52:47 [PhilA]
ack bry
16:53:19 [MikeSmith]
Bryan : We do need to address how [some of this has potential] to break Web applications.
16:54:14 [MikeSmith]
RRSAgent, please draft minutes
16:54:14 [RRSAgent]
I have made the request to generate MikeSmith
16:55:37 [MikeSmith]
Present+ Bruno, Jose
16:55:57 [MikeSmith]
Scribe: MikeSmith
16:56:27 [MikeSmith]
[discussion about communication with IETF]
16:56:48 [MikeSmith]
DKA : Important that we are not trying to invent an adaptation solution in this group
16:57:30 [Rhys]
16:58:18 [Rhys]
16:58:59 [MikeSmith]
Rhys : I think we need to be careful to use phrases like "not aware" ...
16:59:27 [SeanP]
17:00:00 [MikeSmith]
... we have other cases where [because of old server software or whatever] we need for the chain to be robust ...
17:00:24 [Rhys]
ack me
17:00:24 [MikeSmith]
... which is where things like tasting come into play
17:00:39 [MikeSmith]
SeanP : About tasting/differentiating ...
17:01:03 [MikeSmith]
... if we are talking about use of existing stuff like the LINK element [then OK]
17:01:37 [SeanP]
ack me
17:01:56 [MikeSmith]
Rhys : We have added a good number of topics which are now going to be the subject of further work ...
17:02:30 [MikeSmith]
... we have achieved a lot today, as far as what we can manage to get done at this f2f
17:02:54 [srowen]