#social 2017-02-15

2017-02-15 UTC
timbl_, evanminto, KevinMarks, KevinMarks2, cwebber and timbl joined the channel
#
ben_thatmustbeme
csarven: yes, a lot of w3c's specs have basically gone down new roads and not had the implementation experience, but as I understand it w3c has been shifting to being focused on standardizing existing non-standard systems. websub for example, makes a lot of sense in that regard, its been around for years, has had many, many people use it, but it has had holes. It certainly is good to stardardize something that another spec is using normatively,
#
ben_thatmustbeme
implementing they will either abandon using it, or just try solving it some other way themselves, and you end up with non-standard pieces and it would have been better to leave them out of the original spec.
#
ben_thatmustbeme
but if they cannot provide implementation experience back to the source spec, then using the fact that they are normatively referencing as a reason for standardizing it is just building a house of lies. The reason why implementation experience is so necessary is specifically so you keep the spec small and make sure that all the parts actually work. If you have a spec that you releae that has even a few gaps, people will get there when actually
evanminto, timbl and KevinMarks joined the channel
#
csarven
other dimensions into account.
#
csarven
re "house of lies", I think one doesn't have to look far to see that other specs have their own implementations which simultaneously have part implementations of ours. Such implementations are reported in the other WGs and get an approval, however, they are not necessarily reported to this WG. Strictly looking at implementation reports to this WG may overlook other things. The question is not about whether implementations are important or not, but to take
KevinMarks and tantek joined the channel
#
ben_thatmustbeme
csarven: what other specs, as far as I know only annotations has built on top of AS2. but its really not our job to go digging through their implementation reports to decide what is and is not used. thats why we ask for them to submit an Implementation report. You can't say that "because another spec uses it, we assume all things are implemented"
#
ben_thatmustbeme
https://github.com/w3c/test-results/tree/gh-pages/annotation-model has a good number listed there, perhaps email them and ask for implementation reports
KevinMarks and tantek joined the channel
#
ben_thatmustbeme
and while model has a good number of implementations i'm really dismayed that the vocab and protocol are at PR with only 2/3 implementation reports
#
tantek
ben_thatmustbeme: that's not surprising, I think annotations was running out of momentum by the end, and at least they got those 2 (or 3) implementation reports. I think they were trying to do the right thing with requiring 2+ implementations of each feature. I haven't checked details (feature by feature), but I'm presuming good intent based on prior conversations.
#
Loqi
[csarven] could essentially prove the spec's worthiness. Not all tech is strictly here and now, nor meant to be the end of exploration or innovation. A lot of both W3C specs and non-specs out there played a role in building up on new ideas. Sometimes leaving t...
#
tantek
The only thing I would add is that, while I agree with leaving around specs/drafts to help with future exploration, that can be done many ways, as a NOTE, as a CG draft, etc.
#
tantek
And that better helps reflect the status of any such work as of that point in time, so that when people look back at it in the future, they don't get an exaggerated sense of how "solid" it was.
#
tantek
unfortunately we (W3C) have a lot of clean-up to do in that regard, in that there are numerous RECs (in w3.org/TR) which are either obsolete, or were never actually implemented / deployed
#
tantek
and that's not just me saying that, that's a fairly strong consensus discussion / position at the AB and one of the AB's top priorities (cleaning up TR, getting better at spec lifecycle from chartering a a WG to how to do maintenance well and keep specs up to date)
KevinMarks joined the channel
#
bigbluehat
ben_thatmustbeme: fwiw all model implementations are vocab implementations technically
#
bigbluehat
there are other protocol implementations in the works...they just weren't ready within the deadline
#
tantek
bigbluehat what were the requirements for vocab implementations? that was a source of interesting debate here for AS2 (still some lingering questions)
#
tantek
bigbluehat++
#
Loqi
bigbluehat has 3 karma in this channel (5 overall)
KevinMarks joined the channel
#
tantek
got it - we don't have separate exit criteria for AS2 vocab - it has to exit *with* the core
#
tantek
(though in hindsight we should probably have said that explicitly)
#
tantek
the annotation-vocab criteria appear to be focused on parsing / syntax transformation without data loss (which is a good start)
#
tantek
with AS2 we specifically wanted a higher bar than just parsing / syntax transformation, and thus added some wording that implies that: https://www.w3.org/TR/activitystreams-core/#exit-criteria
#
tantek
in particular: "For the purposes of evaluating exit criteria, software that ignores or passes through types or properties inapplicable to their domain have not implemented the type or property. "
#
tantek
we would consider syntax transforms as "passes through"
#
tantek
the intent here being that software has to do something specifically meaningful (like visibly distinct) with different vocabulary or else it has not implemented that vocabulary
#
bigbluehat
yeah. saw that when reading those earlier this week
#
bigbluehat
not sure I agree, though.
#
bigbluehat
sort of breaks extensibility doesn't it?
#
tantek
otherwise it's trivial to just add a bunch of vocab that is roundtrip but doesn't do anything
#
tantek
has nothing to do with extensibility
#
tantek
has to do with the feature set that is exiting CR
#
tantek
any extension specs would have to exit their own CRs
#
bigbluehat
so. that's the trick with validating a vocab
#
bigbluehat
it's a storage format
#
tantek
if that's your assumption
#
bigbluehat
isn't it?
#
tantek
you just transformed the disagreement we have
#
bigbluehat
transformed?
#
tantek
yes - you made an isomorphic statement
#
bigbluehat
yeah me! ;-P
#
tantek
"it's a storage format" is similar to just syntax transformations
#
bigbluehat
so the danger with this requirement "For the purposes of evaluating exit criteria, software that ignores or passes through types or properties inapplicable to their domain have not implemented the type or property. "
#
bigbluehat
is that it would break a baseline intentionally forgiving parser/processor
#
aaronpk
break? don't you just mean wouldn't count?
#
tantek
bigbluehat: not true
#
bigbluehat
I think it comes down to the meaning of "implemented"
#
bigbluehat
and what you're looking for there
#
tantek
bigbluehat: which is why we define that explicitly
#
bigbluehat
oh. checking
#
tantek
first clause of that statement "For the purposes of evaluating exit criteria, "
#
bigbluehat
so. what are they supposed to do instead?
#
tantek
I believe AS2 processors are also required to be forgiving
#
bigbluehat
what would "use the property" mean?
#
tantek
bigbluehat: do something meaningful
#
bigbluehat
one hopes! :)
#
bigbluehat
define meaningful ;)
#
tantek
"ignores or passes through" is meaningless
#
aaronpk
user-visible?
#
bigbluehat
pretty sure I could write a handlebars.js implementation of AS2 renderer
#
bigbluehat
that did all the things
#
tantek
bigbluehat: already did above
#
bigbluehat
but passes through might be exactly what an implementation does do
#
tantek
bigbluehat: see above statement starting with "the intent here"
#
bigbluehat
i.e. aaronpk's "user-visible" bit
#
aaronpk
i would argue if an implementation passes stuff through without "understanding" it then it doesn't really implement it
#
tantek
bigbluehat: passes through is not typically visible, that's just agnostic plumbing
#
bigbluehat
yeah. intentions aren't specs though
#
tantek
bigbluehat: that's the consensus we recorded in our group yes
#
bigbluehat
but maybe the user understands it and cares?
#
tantek
which for purposes of interrpeting CR exit criteria, is good enough
#
bigbluehat
yeah. your call.
#
bigbluehat
i'm not active enough in this group to push back
#
bigbluehat
I just wouldn't want to see AS2 run over at the last minute
#
tantek
again, my preference would be for it to be more explicit in the spec
#
tantek
rather than depending on WG follow-up and resolutions
#
tantek
so yes, we could have done better there
#
tantek
bigbluehat: far from last minute
#
tantek
AS2 was our first WD in the WG
#
tantek
and this discussion about "what counts as implemented" is 2+ years old
#
tantek
anyway, I stated the harm we're seeking to avoid, and I think we're doing that
#
tantek
(not saying annotations has any btw - harms such as that - just to be clear)
#
tantek
(OTOH something like schema dot org definitely has a lot of vocab bloat - very poorly designed at that)
#
bigbluehat
yeah. but it's proof you can be both ugly and popular ;D
#
tantek
bigbluehat: sure, we already knew that from monopolistic examples of the past
#
tantek
or rather, dominant player behaviors
#
tantek
(no need for actual monopoly for such effects to be observed)
#
tantek
ugly and popular only happens if you're already a dominant player
#
tantek
otherwise, you get ignored
#
bigbluehat
k. well, let me know how I can help--especially wrt to getting our Annotation dependent vocab through the flaming hoop at the end of the process
#
tantek
bigbluehat: it's also a bit "early" for schema still. Google has a long history of big giant vocab efforts that eventually failed.
#
tantek
bigbluehat: biggest help would be for annotations implementations to submit AS2 implementation reports ASAP!
#
bigbluehat
which is just via the markdown thingie?
#
tantek
looks for instructions
#
bigbluehat
does ASAP have a point on a calendar at all?
#
tantek
well, for us, the day before our next telcon (2017-02-27) where we assess AS2 readiness to go from CR to PR
#
bigbluehat
oh. looks like Apache Streams got theirs in! https://github.com/w3c/activitystreams/pull/356
#
tantek
I keep hearing there are other deadlines re: Annotations PR->REC that may depend on that
#
bigbluehat
but looks stalled...
#
bigbluehat
oh. strange...
#
bigbluehat
tantek: and yeah, I think our deadline is before yours afaik
#
tantek
I think that's what we're trying to rush for
#
tantek
chairs / contacts are trying to monitor the AS2 impl reports situation for that reason
#
tantek
bigbluehat: to be clear, there's general consensus in the WG to take AS2 to PR ASAP for those reasons, even if that means having to drop unimplemented properties. I.e. we're hoping to have annotations impl reports for the properties that annotations depends on, and other implementations for other properties.
#
tantek
AS2 has already been through a couple (a few? CRs) so there's been plenty of exposure / time on it.
strugee joined the channel