W3C home > Mailing lists > Public > public-rdf-shapes@w3.org > August 2014

Re: is shexc useful?

From: Irene Polikoff <irene@topquadrant.com>
Date: Thu, 7 Aug 2014 09:45:40 -0400
Message-Id: <8FAA21F3-3B3B-49E5-96CC-BB1935877888@topquadrant.com>
Cc: Kendall Clark <kendall@clarkparsia.com>, Holger Knublauch <holger@topquadrant.com>, "public-rdf-shapes@w3.org" <public-rdf-shapes@w3.org>
To: Sandro Hawke <sandro@w3.org>
Great. I am glad we agree.

I think you are exactly right - technology needs to settle down for some periods. During periods of stability adoption grows, then eventually there will be a need for changes to address new ideas and requirements.

I don't see this as a problem. TQ is not concerned about freezing the standardized part of SPIN. If necessary, there is always room to innovate 'on top' of it or in some other, yet to be identified, areas.

And, yes. Perfection is not really attainable. If we get things 80% right in this working group, I'd be happy.

There are some key principles that need to be right. I believe that lack of the wider use of Semantic Web standards, has to do with a small handful of assumptions that ended up being obstacles instead of enablers. Commitment to the OWA to make monotonous reasoning possible, in my view, was one of them. On one hand, this decision has many implications that are too strange for people. On the other hand, monotonous reasoning is only possible when the asserted facts never change, just get added to. In real life, this rarely happens. Data often has errors which need to be corrected, facts that were true at some point get changed, etc. With this, the pluses of OWA don't seem to compensate for its minuses.

As we see with NoSQL movement, there is a big need for a flexible data model. RDF provides that. Having a standard query language in SPARQL is great.

It is excellent to finally see a working group addressing data definition and validation needs. This has been a missing piece in the standards.

Irene

> On Aug 7, 2014, at 7:06 AM, Sandro Hawke <sandro@w3.org> wrote:
> 
>> On August 6, 2014 4:51:21 PM EDT, Irene Polikoff <irene@topquadrant.com> wrote:
>> Sandro,
>> 
>> 
>> 
>> 
>> 
>> < That's essentially true of every technology working its way through
>> W3C.  Whatever comes into the process gets turned into something else
>> on its way to Rec.   When things work properly, along the way, it picks
>> up lots of production implementations, gets lots of companies standing
>> behind it, and probably gets some books.   Whatever of those things it
>> had before starting the process.... those things maybe help bootstrap
>> the process, but they're also a drag on the process since they often
>> require backward compatibility.    Are the current SPIN and ICV users
>> perfectly okay with those technologies being changed by the WG?   So
>> having those things at the start is a mixed blessing.   Probably a net
>> positive, but still a mixed blessing.>
>> 
>> 
>> 
>> On this topic you and I have an important philosophical disagreement. I
>> believe that practical experiences of using implemented technologies in
>> deployed situations (and I don’t include academic work in this
>> category) are critical to “getting things right”. They keep one
>> grounded, focused on what is real and what is important as opposed to
>> someone’s theory of what may be needed.
> 
> No disagreement at all.   I completely agree that practical experience is tremendously useful.   I might argue it's never quite possible to "get things right".
> 
> I guess there are two issues at play.
> 
> 1. The process of getting buy in from a wide community seems to nearly always involve allowing members of that community to invent new things.  I'm not saying this is good, but it ends up often being necessary.  Key vendors always seem to want to tweak something.    Maybe it's because they just see the world a little differently, or maybe it's something else.
> 
> 2. A fundamental part of the problem is that every technology needs to evolve to match a changing world and a changing understanding of the world, and yet in order to have very wide adoption a technology needs to be frozen.    So it inevitably is frozen before some important ideas or changes to the world.
> 
> In html5 there's a split in the community which I don't fully understand but I think is between the folks who think they only need buy in from a handful of browser manufacturers, so it's easy to keep improving things, and the folks who are aware of a vast ecosystem around html and see a need for punctuated equilibrium, where things settle down between periods of change.
> 
> Holger mentioned tq sometimes tweaks SPIN.   That's going to become awkward if SPIN or part of SPIN I'd standardized.
> 
> Anyway, those are two issues I wanted to point out, but I certainly agree it's good to stick very close to something that's well known and well liked.
> 
> Hopefully the whole WG can get to know all the options before making a decision.   But you're right that things with a longer history and more users can be much better known.
> 
> 
> Without this, having the
>> resulting standard “work properly” is very unlikely. Does it happen?
>> Yes, but not often.
>> 
>> 
>> 
>> When a new “standard” reaches a recommendation status, there will be
>> some people implementing it, some people writing about it, etc. whether
>> it is practically useful or not. But, if it misses the target, this
>> community will remain a very small minority. Unfortunately, this has
>> been true for a number of W3C recommendations.
> 
> Indeed.  In my less charitable moods, I'd include RDF, OWL, and SPARQL.    They have a large number of users, but I think still have a tiny fraction of the potential users.
> 
>    - Sandro
> 
>> 
>> 
>> Regards,
>> 
>> 
>> 
>> Irene
>> 
>> 
>> 
>> 
>> 
>> From: Sandro Hawke [mailto:sandro@w3.org] 
>> Sent: Wednesday, August 06, 2014 2:38 PM
>> To: Kendall Clark
>> Cc: Holger Knublauch; public-rdf-shapes@w3.org
>> Subject: is shexc useful?
>> 
>> 
>> 
>> [changing the subject, since you changed the topic]
>> 
>> On 08/06/2014 08:58 AM, Kendall Clark wrote:
>> 
>> On Wed, Aug 6, 2014 at 8:30 AM, Sandro Hawke <sandro@w3.org> wrote:
>> 
>> 
>> Is it?    To me it looks like someone combined data structure
>> definitions from any language that has such things (Pascal, C, Java,
>> Go, ....) with the Kleene operators, known to every programmer from
>> EBNF and RegExps.
>> 
>> Eric may have been thinking about relaxng, but the design makes prefect
>> sense and seems completely familiar to some of us not steeped in
>> relaxng
>> 
>> 
>> 
>> Isn't this still a vendor organization?
>> 
>> 
>> W3C is probably more accurately described as a multi-stakeholder
>> organization.   Its mission involves the good of the Web, not
>> specifically the good of its members or a set of vendors.   The
>> arguments that should win the day are grounded in what's going to be
>> long-term good for the Web (and the world at large), much more than
>> good for one specific vendor or set of vendors.
>> 
>> 
>> 
>> 
>> 
>> Whatever you want to say about the origin and quality of ShEx syntax (I
>> don't like it at all, personally), the fact remains:
>> 
>> 
>> 
>> * it has no users
>> 
>> 
>> The reply from Jose Emilio Labra Gayo shows otherwise.
>> 
>> 
>> 
>> 
>> * it has no production implementations
>> 
>> * it has no *company* standing behind it to support it in the market
>> 
>> * it has no experts writing books about it
>> 
>> 
>> That's essentially true of every technology working its way through
>> W3C.  Whatever comes into the process gets turned into something else
>> on its way to Rec.   When things work properly, along the way, it picks
>> up lots of production implementations, gets lots of companies standing
>> behind it, and probably gets some books.   Whatever of those things it
>> had before starting the process.... those things maybe help bootstrap
>> the process, but they're also a drag on the process since they often
>> require backward compatibility.    Are the current SPIN and ICV users
>> perfectly okay with those technologies being changed by the WG?   So
>> having those things at the start is a mixed blessing.   Probably a net
>> positive, but still a mixed blessing.
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> I have told Holger many times privately that I don't really like SPIN
>> too much, but it has *all* of the above things and more. Same goes for
>> IBM Resource Shapes.
>> 
>> 
>> 
>> ShEx is a research project and nothing more. I thought we weren't doing
>> R&D in W3C WGs any more?
>> 
>> 
>> 
>> 
>> It seems to me like it's aways a matter of degree, with every WG doing
>> some R&D.   What we don't want is a REC whose predicted success is
>> based on untested assumptions.    But all these inputs took to me to
>> have various untested assumptions, when you consider them being applied
>> to all the use cases being presented.
>> 
>> 
>> 
>> 
>> 
>> There are *three* at least *adequate* commercial solutions to start
>> from. There's simply no need for ShEx.
>> 
>> 
>> 
>> 
>> I've seen a variety of statements from people on this list, and at the
>> workshop, that seem to disagree.
>> 
>> 
>> 
>> 
>> Cheers,
>> 
>> Kendall 
>> 
>> 
>> 
>> PS--No offense meant to EricP: he's a fine researcher in this space and
>> I'm sure there are several good papers to be written about ShEx. 
>> 
>> 
>> My understanding is Eric isn't trying to do research.   He's been
>> trying to solve the problem that the Workshop brought to light. 
>> Success will be technology adoption and user satisfaction, not
>> publications.   If the technology turns out not to be useful, that's
>> also success for Eric -- the problem is still solved.
>> 
>> 
>> 
>> 
>> But this is standardization of a space that has, in some sense, *too
>> many* starting points, not too few. SPIN, Resource Shapes, and even ICV
>> are all *real* systems in comparison.
>> 
>> 
>> Yes, but the workshop conclusion was none of those were sufficient.  
>> Perhaps that conclusion was wrong, I know.   This is something the WG
>> will have to determine.   IMHO there's no point is us trying to figure
>> it out via this list.
>> 
>>    -- Sandro
> 
> 
Received on Thursday, 7 August 2014 13:46:14 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:02:40 UTC