- From: Dael Jackson <daelcss@gmail.com>
- Date: Mon, 30 Sep 2013 19:58:33 -0400
- To: www-style@w3.org
- Message-ID: <CADhPm3uf47Fa3+s7B7=r42ehm_bowQ-tSiUyN2OYE6OgsUe3Cw@mail.gmail.com>
Charter
-------
- There was discussion about how to handle timelines for specs as a
part of the rechartering process.
- The group also discussed possible alterations and alternatives to
the current charter model that may better fit the process of the
WG.
- It was decided that the chairs, plinss and glazou, would draw up a
list of priorities from existing WG documentation and submit it
to the WG for review and approval before inclusion in the charter
documentation.
=====FULL MINUTES BELOW=======
Charter
-------
Scribe: fantasai
<Bert> http://wiki.csswg.org/planning/charter-2013
<dbaron> I disagree with the "what people expect to see in the
charter".
Florian: It's easy to say as an editor that will make progress, hard
to say it will definitely will make it to transition to next
status
<dbaron> I don't think this exercise is useful.
Bert: This was starting point for discussion
<astearns> dbaron: do you have an alternative exercise in mind for
preparing our charter?
<dbaron> astearns, not listing fake deadlines for specs?
fantasai: What are we trying to come up with to put in the charter?
Bert: Most charter's say how long it'll take to get x to y status
Bert: Our charter has traditionally just listed what status we expect
to have at the end of the charter period.
fantasai: I think we should just go with the format we used last time.
ChrisL: I did a team-internal study on why specs were late
ChrisL: And found out that it was largely because people were
pressured to put deadlines of 6- months to 2 years
ChrisL: Because of afraid the charter will be rejected
ChrisL: If they gave the real answer.
glazou: We have 2 AB members in the room.
glazou: I want to say something;
glazou: CSSWG is the oldest WG, rechartered every 2-3 years.
glazou: I think system of chartering and rechartering is wrong for
such a group.
glazou: Charter's are great for group that is to do a specific thing
and then close,
glazou: But we don't do that, we start things that go on, and start
more things that continue after.
glazou: We spend a lot of time on chartering and rechartering.
glazou: So I'm ready to take an action, this is something I'm saying
to AB members, I want to send message to AC forum describing
the issue,
glazou: And asking if W3C staff would be willing to find a better
solution.
SteveZ: 3 answers:
SteveZ: It's already under AB agenda under title Super-groups.
SteveZ: That is, groups with multiple work items progressing largely
independently. Such as webapps has similar concerns.
SteveZ: We don't have a good solution to the problem yet.
SteveZ: The problem you just described is recognized and on the
agenda.
glazou: Do you think email describing issue can help?
Tantek: Yes.
SteveZ: Yes.
SteveZ: Fly in the ointment is basically the patent policy.
SteveZ: That's because what's on the charter is what people are
committing to potentially make IPR commitments too.
SteveZ: So for that reason updating charter on regular basis is not a
bad thing, I think.
SteveZ: It's every 2 years.
SteveZ: Yes, takes a lot of time, but not all that frequent.
SteveZ: But it forces us to do the evaluation, which we maybe ought to
be doing more frequently, of are we making the progress we
thought we'd make 2 years ago, and if not, why not.
SteveZ: It's a useful exercise in that respect.
SteveZ: Most companies do it every year.
Florian: I think putting dates on things is sometimes useful.
Florian: But if we do it for everything, even when not useful, it will
confuse things.
Florian: For example, if we had charter that said IE 10 will release
by date X, and that Flexbox must be done before that, it's
useful to say so in charter.
Florian: Useful to say we want to put in date because the date is
important,
Florian: But if there's no reason for the date to be important, then
we shouldn't put it down.
tantek: Would it help to send an email, the short answer is yes. If
you're ok with it, I would also ask to CC www-archive, because
I think this is a reasonable discussion to stay public.
tantek: On specifics of what you raised, I think you'll find as
someone who also likes less process and less bureaucracy, I
would agree with most of what you said.
tantek: The act of rechartering, for CSSWG, is here are list of things
we want to work on, with approximate priority for next N
years. I think that's useful.
tantek: I think that helps communicate to outside world what WG
considers important, and gives the outside world the ability
to communicate back if they disagree.
tantek: So I think that's useful.
tantek: Sometimes we may want to update that more often than just
every 2-3 years.
tantek: About the dates of delivery, I think evidence has shown that
any prediction is futile; I have yet to see of them hit by
anyone.
tantek: So I will counter florian's assertion about dates
tantek: with saying that if any editor wants to promise a date by
which they will deliver a certain draft level, they're welcome
to do so,
tantek: But to ask the WG as a whole to do so, is ridiculous and
bordering on dishonest
tantek: Based on how none of that has ever worked.
tantek: But if there are editors who feel confident in their ability
to predict deadlines, go for it.
Florian: I'm not sure we're actually disagreeing.
<sgalineau> Was a *promise* requested? or an *estimate*? those are
very different things
glazou: I agree entirely.
glazou: I'm making this discussion diverge a bit to meta-problem
glazou I really feel we have an issue. This is 3rd time we are
rechartering.
glazou: Every time it took a few months,
glazou: Once it took 6 months,
glazou: And it sucks up our time.
glazou: Since AB works on that, I didn't know that.
glazou: Maybe we could we serve an experiment?
glazou: If you're looking for an example if we can make an example,
could the CSSWG be that example?
glazou: And could we write a charter that lasts a long time, and be
amendable in a very simpler way.
glazou: Amendable by note.
glazou: I agree that the deliverable sections and milestones sections;
we never meet them, they are crazy.
glazou: The potentially expected deliverables and milestones are
just...
glazou: We can reach the CR status, CR and rec is in between
implementers and writers out of tests suite and not fully in
our hands.
glazou: If the AB is interested, for this rechartering, in another way
of doing a charter and managing a working group, even with the
patents issues and everything
glazou: I would really be glad.
tantek: I think that's somewhat in your hands. AB is meeting soon.
tantek: And make a proposal per email as said earlier
ChrisL: Another thing that used to happen, the AC would worry about
how many groups work working at one time.
ChrisL: So they would charter group to do specific thing, and cease to
exist.
ChrisL: Then we had idea of adding 6 months to the end to do
maintenance work.
ChrisL: Then we also found out later that if you closed down a WG, and
then reopened to do second version later, you lost all the
patent commitments.
ChrisL: It didn't work for short WGs so much.
ChrisL: MathWG oscillates between doing something, and then sits there
for 2-3 years so it doesn't go away.
ChrisL: Whether there's multiple specs or one, the assumption should
be that to keep all of the IP commitments alive, the group
must not close,
SteveZ: Daniel, your comments you hit on the key point, which is
management
SteveZ: of the WG.
SteveZ: There's 2 organizations at least that care about that, one is
W3C Team, and other is AC
SteveZ: I agree with Tantek's comments and Florian's comments that the
value of dates is in the short term and not the long term.
SteveZ: But those dates were what the team was trying to use to see if
progress is being made
SteveZ: So any proposal to eliminate them has to come up with a way of
indicating progress.
glazou: Yes, I understand fully
glazou: It's a way of simplifying things, not eliminating control
Florian: I think I was disagreeing a lot less with Tantek than he
thought.
florian: If we have a commitment for a date, and I expect this to be
almost never or never happen, it's useful to state.
florian: E.g. in Flexbox case, there was sense of time period.
florian: If the working group is not committed to that time, then it
works to put that date
* sgalineau animations is only two years behind my first CR estimate
*cough*
tantek: I think time urgency should be driven by implementers
hober: So, ChrisL described previous model of spinning up and shutting
down groups which evolved into a hybrid of hibernating and then
rejuvenating groups.
hober: For core platform work, that model is fiction.
hober: We have to be continuously working on core platform.
hober: Steve described effort of rechartering as like annual reviews
in a company
hober: I think it's useful to have annual review.
hober: However, I would like to decouple that from rechartering.
hober: HTML, CSS, WebPapps need indefinite charters.
hober: As mentioned, going from CR to REC is not in the WG hands
hober: Now that we have core testing effort happening at W3C, I think
one suggestion we can make at AB/AC is to make these CR
transitions to be joint effort of testing efforts and that WG.
hober: It's partially out of our hands,
hober: Let's acknowledge that.
plinss: Having been in core testing group, they really don't give a
shit about advancing specs along REC track.
plinss: The folks involved in web platform testing effort are not
interested in advancing REC. Just interested in improving
quality in general.
plinss: There is a very small subset that does care about moving to
REC. Not my experience that that has any core priority.
plinss: "If we can do both great, but we're not going to worry about."
plinss: That's kept me from committing to working with their
infrastructure, because if I say "we need this to advance
specs to rec" they say "we don't care about that."
hober: Going from CR to REC should be joint effort, and apparently
feedback from testing effort is not interested in that.
hober: Unilateral option is that it still is a joint effort, and just
make our primary goal reaching CR.
Liam: We need to separate possible procedural changes in the future
and the next rechartering, and to note the charter does not
commit dates.
Liam: Any change to W3C chartering process is not overnight, will take
year or two. So be prepared to do current re-chartering as
previously.
Liam: However, a common misconception among WGs, including Team
people, is that the dates in charter are a commitment. They're
not. They're just to document deviations in charter that will be
documented on homepage.
glazou: Kind of a thing in W3C Process that nobody reads/cares about
<astearns> so we charter with 2yr dates, then we immediately document
that those dates are meaningless.
dbaron: I think it's worth discussing relative priorities of specs,
and don't think it's worth discussing dates.
dbaron: If we need to put dates, someone should just make them up.
dbaron: I hope that it doesn't, and can just be left out.
glazou: With regards to changing the charter process, I'm happy to
just ask for a charter extension. If an experiment in 6 months
is feasible, I'm willing to try that.
glazou: Every time we recharter, has been long and painful effort.
glazou: If there's a better way to do it, I'm willing to try
zcorpan: Different people have different goals. Some people want to
advance to REC, and some people want to find as many bugs in
browsers and get them fixed.
zcorpan_: The testcases in those two goals, have different incentives
for writing tests
zcorpan_: REC people look at a test and their incentive is that it not
finds bugs because it blocks the spec.
ChrisL: Another difference is that platform tests tend to test
interactions among tests, whereas REC tests focus on single
spec. There's definite need for interaction tests.
ChrisL: We also need interaction tests within this group
ChrisL: We're not testing that, but it's important.
ChrisL: That could show us problems in the specs, that you could only
see in combination of things.
glazou: So I think we're agreed that first of our priorities in
rechartering is defining list of priorities.
glazou: Figure out what we want to work on. Milestone section is lower
priority, work on list of documents first.
florian: On the topic of tests, I think the behavior zcorpan talks
about is something we did.
plinss: I always describe our testing as 2-phase. Spec testing and
conformance testing.
plinss: There's no reason why we shouldn't build both test suites in
parallel.
Florian: If you go for one goal, you're not going for the other.
plinss: Want to build both test suites in parallel.
plinss: Say these are our spec tests, and these are our conformance
tests.
tantek: I agree with prioritizing the prioritizing, in general
tantek: But I think if something is not important to us, we should
drop it regardless of what's required by process and charter.
tantek: Drop sections of charter that we don't care about
tantek: And move forward with that.
tantek: Make that proposal, and I'll argue for that before AB.
glazou: The history of WG has shown that we're really bad at
estimating time for a spec, but we eventually finish it.
glazou: So there are usually specs on the WG radar and they stay on
radar until they're done.
SteveZ: Requirement is expected milestones, meaning we don't expect
any, then we don't need to record any.
ChrisL: One reason for this scrutiny is because some groups, including
in the past this one, have spent 10 years not producing any
RECs.
ChrisL: Since then have been producing things regularly/
ChrisL: So we can point to that track record.
dbaron: The group wasn't sleeping.
ChrisL: From the outside, you couldn't tell that it was doing
anything.
glazou: The web designers community was really mad at us, because we
*seemed* to be doing nothing.
ChrisL: We were doing a lot of detailed work on CSS2.1
ChrisL: But from management POV, seemed like we weren't producing
anything.
[Break for Lunch]
glazou: Discussion diverged a bit from original goal, but
meta-discussion probably is over.
glazou: We need to reach a list of priorities
glazou: Still have option of doing what we did few years ago, asking
vendors to send list of priorities
fantasai: We did a poll recently, no?
glazou: It's kind of old.
glazou: Or we could review list of documents now.
fantasai: I don't think it would be too bad to put together a list
right now, based on the old data
fantasai: Doesn't seem like there's anything controversial in the
group right now.
glazou: So maybe we take an action to draw up a list and ask for
group's feedback.
glazou: So what do we do with milestones section?
fantasai: I suggest we follow Tantek's suggestion and leave them out.
Replace it with pointer to current-work page.
tantek: Since this group has gotten better at keeping a list of
priorities for specs, maybe it's not worth group's time to
discuss it in person.
tantek: I would trust the chairs to take existing list, make any small
adjustments as they see needed, and send that to group for
review.
tantek: It doesn't need to be discussed in a f2f. It is fairly
uncontroversial and doubt we'll see much controversy.
tantek: So let's move forward with that optimistically.
glazou: Other opinions? Or +1? or -1?
<dbaron> +1
<florian> +1
glazou: Ok, we'll do that.
glazou: Related is EPUB interest group.
Bert: I think that's less important than list of priorities.
Bert: But we need to write something in liaison section.
Bert: Do we want to have any closer cooperation with this group?
glazou: Anyone from interest group that is member of this group?
Bert: Hachete
Bert: Peter
Bert: Bloomberg
Bert: Don't think anyone in room, aside from Peter
fantasai: I think we can figure out logistics of it later, not
necessary to put in charter.
fantasai: Just put that we will have liaison
ACTION plinss, glazou: List priorities in charter, submit to WG for
review and approval, then submit to staff contact for proposed
charter.
ACTION glazou: Email AB with regard to rechartering woes
glazou: Ok, we still need jdaggett for Text
Received on Monday, 30 September 2013 23:59:01 UTC