Future of W3C publishing process -- 31 Oct 2012

<Lachy> ScribeNick: Lachy

<plh> http://www.w3.org/2012/Talks/1031-tr-plh/

Future of W3C publishing process

<plh> http://www.w3.org/2012/Talks/1031-tr-plh/#(1)

<betehess> scribenick: betehess

plh: this is about the future of #TR
... not about the process itself
... only about /TR pages
... and how to publish stuff there
... not about new style either

Lachy: so this exlucdes stuff re: WD, LC, etc.

steeve: actually, I'll be happy to talk about that after the session

<cygri> Richard Cyganiak, DERI

Alexandre Bertails, W3C Systems Team

plh: my goal is to get feedback
... identify what we need to do
... go away if you don't know what /TR is

<SteveZ> SteveZ is Steve Zilles

plh: http://www.w3.org/TR/ is the canonical URL
... this has evolved over time

<Josh_Soref> THE FUTURE OF /TR

plh: list technologies, status, date, title, etc.

<Josh_Soref> WHAT'S FOR TODAY?

<Josh_Soref> WHAT'S /TR?

plh: provides several views to access specs

<Josh_Soref> WHAT'S /TR?

plh: Submissions are not there, like other docs

<annevk> Feedback: in the Netherlands the last name "Van Kesteren" is sorted using K, not V

plh: the W3C publication process is a bit complex

<annevk> (Probably easier to sort on first name.)

plh: reflects the w3c process
... sometimes you need to ask permission for transitions.
... then you ask the webmaster for actual publication
... he can say no at any point if doc is not ready
... the webmaster also deals with the comm-team
... to make sure they are ready
... then he interacts with very old back-end
... then it's finally published
... in the meantime, the comm-team does announcements
... eg. twitter, the AC, etc.
... may depend on the kind of document as well
... as an editor, it's not easy

<Josh_Soref> PUBLICATION PROCESS AS OF TODAY (SIMPLIFIED)

<Josh_Soref> WHAT'S IN /TR?

plh: I'll focus on informations in /TR
... "this version" is immutable
... a link to the doc itself
... "latest version"
... "previous version"
... and now, new stuff, like "editor's draft"
... (not required)

<Josh_Soref> WHO USES /TR?

plh: who uses /TR, the web page?
... do you go through this page to find stuff?
... or maybe a search engine?
... what about webdevs?
... and sometimes, you want to know about other resources
... my question: do you use it? how? why? if not, why?

Who uses /TR

Lachy: I use Google
... just easier
... I usually don't know what I'm looking for
... there are too much informations on the webpage

<sandro> Lachy: Sometimes I know the shortname, sometimes it's in my browser history

@@: also dev.w3.org

scribe: because I want the latest editor's draft

<sandro> hsivonen: I add site:dev.w3.org if I what the ED.

Lachy: depending on the spec, I may want the latest draft, or something in /TR

Travis: some years ago, I was going to /TR
... to see waht was done
... as I didn't know the spec
... don't do that anymore
... I want to have the latest version

<sandro> Travis: I want the ED, so I dont use /TR

vivia: if I want to find information, I use google
... but sometimes people point me to the editor's draft

david: I used to use the /TR page as a summary
... but can't find things in there anymore

<sandro> +1 dbaron I stopped using /TR when it became dynamic

david: stop using it
... too dynamic
... so I'm using google now

pl: I don't use /TR either

dom: I use it sometimes

chaals: I use Yandex now :-)
... when I use it, it's when I want to give people a link to a handful of specs
... like an overview
... and for history stuff
... I can see things by date order
... and I point people there when they have questions

<sandro> chaals: one use is to point people at the progress of a group, or a set of related specs

chaals: eg. they want the REC version if there is one

plh: what about HTML?

chaals: depends on the issue

<sandro> chaals: I used autocomplete for the URL for the HTML5 draft

chaals: most of the time, it's dev
... sometimes /TR version

<SimonSapin> (I use the single page HTML spec as a crash-test)

doug: some people told they always use editor'ED for tests
... also, we have more ED than REC from the beginning

cygri: I've used it twice in my life
... to find what other specs where using RDF

<annevk> More feedback: lots of people don't know what "TR" means

cygri: but still hard to figure out where the information is
... can't find the right categories
... 2nd time, when I had to publish a FPWD, I had to find some extra information, like the group

darobin: I use for one thing
... I'd really like a separate version of this page for ED
... machine readable, not RDF please

fantasai: it's not useful at all
... not sure we need a separate page
... it should be only about ED

steeve: if I'm a user, do I want that?
... things may be inconsistent
... people in this room are talking to a particular audience
... the audience is probably not the people in this room

doug: developers should not read the specs

<annevk> angry mob

fantasai: would to clarify
... speaking about the version the WG considers the people should look at is on /TR
... for other WG, this is different

<sandro> +1 fantasai: the version on TR should always be the version the WG thinks people should be looking at

<Josh_Soref> [ applause ]

fantasai: we should be able to point people at where the WG thinks they should look at

<Josh_Soref> /TR GOALS

plh: /TR is only for snapshots *today*
... (approved by WG)
... other possibility, the latest version could be the thing that could be modified at will
... but when I'm in LC, which one should I look at?
... things can be moving

sandro: we could point people at the snapshot, not the latest version

plinss if it's the LC, it cannot be a moving target

scribe: people should not look at it

marcos: what's LC for?

plinss the process

marcos: does not have to be the stable doc
... only the lawyers need that

steeve: two uses for LC

<dbaron> annevk: You don't want to get the same feedback multiple times when it's been fixed.

steeve: 1. is for patent reviews
... to see what can apply
... requires stable doc

<sandro> +1 annevk: If part of the comments are already addressed in the ED, you don't want people looking at the old text.

2. is for people to have something to review

scribe: and make comments
... they are interested in the conclusion
... not the moving thing
... snapshots are intended for people not in this room

<annevk> hmm

ArnaudLH: wasting our time
... people should be given the choice
... we should be telling people the right thing
... we just not choose for them
... should be easy

<sandro> +1 ArnaudLH just provide all the information, and let people pick which version(s) they want to see

hsivonen: don't think it's only for lawyers
... if you write a comment, just look in the spec if it's still accurate
... you can remove your comment
... so snapshots are not useful
... if you give choice, there is a risk that they'll read the wrong one

marcos: we can look at the evidence
... it's never been outside of the community

<sandro> hsivonen: If you give people a choice, then you run the risk that New Hires will read the wrong one

marcos: I bet that comments are always coming from the community

dbaron: @ArnaudLH giving people the choice, URLs are passed around, and people may be pointed to the wrong spec
... but I agree, we should be able to give the option
... but people link to the specs all the time
... @Henry, there is more value in snap than for lawyers

<sandro> that sounds like a small matter of programming

dbaron: when reading/writting comments for specs
... as I want to point to a specific version of the spec
... wants to speak about a specific paragraph

robin: don't think that snap are only for lawyers
... I get feedback for implementors
... so snapshot are useful
... I like the idea we have to generate several versions

<annevk> Snapshots don't have any better stability in terms of feature review...

<annevk> You need annotations for feature stability

robin: we could make it so we detect if we have a fresher version of the spec

Lachy: to dbaron re: linking to specific version, revision number from VCS in the URL is enough

<Marcos> +q

Lachy: @SteveZ ... as an implementor, we have different people working on different sections of the spec
... does not matter if it's LC for them, it's not relevant
... in this case we ignore the fact it's LC

SteveZ: I accept the comment re: LC doe snot mean it's not finished
... but in this case, could mean that you're not really in LC
... specs are primarly for implementors
... so the URL should be the most useful to them
... could be reasonable to provide URLs to derivative
... still, we should provide snapshots for people who want them

doug: the specs are hard, not for newbies

<timeless> scribe: timeless

Travis: thinking about fantasai's place
... having a place which is semi private

<sandro> "development branch"

fantasai: a place where i can discuss on the ML

Travis: a place where she can have implementers try it out
... we've made ED ....

darobin: half baked, confused

Marcos: i'd like to rename LC to Lawyer-Call instead of Last-Call
... if we're always in a state of receiving comments

annevk: we can keep "LC"

darobin: some groups should randomly go to REC

jgraham: if you want a private scratch place
... it sounds like you should use branches in VCS

sandro: i've always wanted to make publishing a WD a one-click
... so ED appears on TR as a WD
... maybe Editors or Chairs clicks that button

plh: nice transition to my next slide

fantasai: i'd probably like to push all typo fixes and similar fixes *immediately* to TR

<sandro> with "minor changes" check box

plinss you want the "minor changes" checkbox

fantasai: checked by default
... in the CSS WG
... the process of requesting a publication from resolution
... causes people who were procrastinating to look at this NOW
... so there are aspects of publishing a snapshot for every change
... i want /TR up to date enough that people don't feel the need to look at my scratch

<sandro> maybe "editorial changes"

darobin: +1

<Josh_Soref> PUBLICATION PROCESS

plh: is the current process painful for you?
... is pubrules painful?

[ laughter ]

plh: referring to sandro
... how often
... i've heard w/in 5 minutes

ashok: i work w/ other SDOs
... I work w/ OSS
... UMTF
... with those guys, you author a document
... in Word, PDF
... and upload it
... with W3
... it takes me as long to publish as it takes to write it

[ laughter ]

ashok: i can't publish what i edit
... it has to go through what i process
... there's this CVS business
... why is this that difficult?

wilhelm: at the current stage
... webdriver is in
... i want a commit hook
... editor commits
... we never use /TR
... what's there is wrong

dbaron: i'd like to offer some data

plinss: 1000010000

[ laughter ]

dbaron: longest period of time from a WG Resolution to publish a doc on TR
... to actual publication is 86 days
... only the ones i've been involved in
... Apr 24, 2007, to July 19, 2007
... the longest in the past 6 months is 50 days
... Aug 1, 2012 to Sept 20, 2012
... WG Res to Publication
... this is not abnormal

<dom> [what were the causes of the delays, though?]

ArnaudLH: i think pubrules are not too painful
... i'd like to publish as often as i'd want
... i want to completely control publication
... not have to try to get team contact/webteam/sysadmin
... i want a button
... make it happen in a few minutes

SimonPieters: my comment is the same as everyone else
... publication process should be completely automated

doug: so should the authoring process

Lachy: re pubrules
... it requires a lot of edits to update the little things
... to go up a stage
... i have to change stylesheet, link to previous
... that takes a lot of time
... updating SoD

darobin: are you using Anolis?
... that's not a tool

Lachy: manual edits is annoying
... should be able to say Spec @ Level - Go

plh: not many of us use /TR
... since it became dynamic
... want to find latest version agreed by WG
... some mixed opinions, but WG agreed latest ver is the one to show
... dbaron mentioned link which is flexible to recipient

<dom> [ReSpec is one possible part of the solution to the problem "we want to automate publication"; that doesn't invalidate the expression of the need]

plh: one click button

fantasai: commit hook

plh: i'll stop

SteveZ: announcements only w/ identified major revisions

<dbaron> ... not with publications

Josh_Soref: and if there's some final minor update to a major version (HTML4.01) it should have a publication announcement

- DRAFT -

Future of W3C publishing process

31 Oct 2012

Attendees

Contents

Future of W3C publishing process

Who uses /TR

Summary of Action Items

Scribe.perl diagnostic output