[OSGeo-Discuss] Re: [OSGeo-Standards] TMS and WMTS

[Brian Russo]

On Wed, Apr 7, 2010 at 7:10 AM, Seven (aka Arnulf) <seven at arnulf.us> wrote:
>
> Hey,
> sorry to spam Discuss with nerdy smalltalk. We might want to move back
> to the standards list for follow ups.

I've moved it to standards (which I wasn't a member of.. so now I am).

>
> Brian Russo wrote:
>> I have a simpler/better idea - have the OGC stop creating
>> unnecessarily complex standards hundreds of pages long that hardly
>> anyone implements. This will save time/money, and benefit users,
>> proprietary and open source developers alike.
>
> Yes making standards readable and usable is a great idea. It takes time
> and brains to implement a good standard. I lack both, so not a good
> choice. How about you? If you are good at this why don't you join the
> OGC process and help do it better? But watch out, the OGC has a high
> frustration potential because there is always knowledgeable folks around
> who pick apart what you just put together. Which is why some standards
> actually work pretty well.

Honestly I have never even considered this and I think that is a
problem organisations like OGC face. The people actually doing this
stuff are too busy .. actually doing stuff to write about it. So what
happens is you get a lot of academics and industry people that develop
these standards. Now I'm not trying to slight them - their
perspectives are great. But they often miss out on a lot of the
'reality' side in terms of the real world issues we face. It's not
even so much that they won't consider the issues, but the weighting
may be different. So while for example to some guy with a 1 GigE dev
lan bandwidth constraint is not an issue. Well to me running forward
operations with an unreliable 500Kb pipe - well yeah that is a big
concern for me. And so on..

As far as contributing, sure I'd be interested in developing
standards. And I'm not saying it isn't hard. Doing anything right is
usually more difficult than doing a half-assed job - that's why
there's so much crap in the world. Frankly I'm more of a meddler which
is why I think do pretty well in my current job.


>
>> Sometimes I think it's a concerted effort to make sure the 'open'
>> standards are as complex as possible so few people have the resources
>
> Thank you for the laugh but you do not believe this yourself. So why say
> it? This is exactly the tone I regret in this discussion. OGC is neither
> a conspiracy nor are they all brain dead. I might be both, conceded, but
> this is beside the topic.

Maybe the sarcastic tone is a bit counter-productive, but in a sense
it is true. We talk about 'openness' and 'transparency'. But you can't
just build it - you have to sell it. And no I don't think people are
stupid so often as they are merely approaching issues in too much of a
focused and not a holistic enough perspective. So if the 'product'
here is a standard, then it's not enough to have an open standard. It
has to also stand on its own as a good one and by the way - you have
to sell it. Part of selling it is making sure that people can actually
implement it and want to.

So while I appreciate Bruce's comment that nobody is forcing me to use
open standards - sure I got that.. but it's not about who's forcing
whom. Our standards should be so awesome that people are throwing
themselves at us saying 'hey, how do I get your standard into my
product? It's in such huge demand!'


>> to implement them (except proprietary vendors and academics with tons
>> of time) and the rest of us all stick with proprietary standards
>
> I guess that you will be of the same opinion as me that an open standard
> that all can use for free is better than a costly and potentially patent
> infected proprietary standard that can be changed at the whim of its
> singular (proprietary vendor) owner. At least I can see a difference.

Yes.
I can certainly see a difference and in no way am I supporting
proprietary standards. I am hugely in favour of open standards -
because they ultimately drive open source software.
But above all I am in favour of good standards.

And it's not so simple for various reasons (technical, cultural, etc)
to just one day snap your fingers and transition over to everything
being open. I am envious of people that do that but.. especially in
government it's just not going to happen. It doesn't mean you give up
improving things - but you can't do it all in a day.

The first step in improving business processes & information flow in
an organisation is to reduce the number of data formats and make those
formats more open. This is true regardless of whether you're talking
financial documents or spreadsheets.. geospatial.. whatever.
Ultimately that makes it possible to use open source software. People
forget that the goal is really the information/content itself. I don't
care about the system it 'happens' to be running on. The
standard/format is merely a vehicle that holds my content and lets me
share it to the world. Open formats free information. Closed ones..
lock it up. But an overkill standard never gets implemented in the
first place and isn't even in the running.

>
>> (because we have the software - the lazy solution), or simple open
>> ones like GeoRSS-Simple (because a normal person with a normal
>> schedule can actually understand it).
>
> That is another one that should go under the hood of OSGeo, if it is to
> become of any relevance for example to INSPIRE. If they want to use it,
> it has to become an ISO standard, else you can't stuff it in a law.
> Stupid, aint it?

Just looking at geodata.. the standards we use on a routine basis
include WMS, VPF, ESRI Shapefile, ESRI 9.2/9.3 REST/SOAP services,
KML, GeoRSS and GeoJSON. How many of those are OGC standards? Only 2
as far as I can tell (KML/WMS). Do we have too many? Maybe. Ultimately
it doesn't matter if my information (remember.. it's about the content
not the container) can freely flow from 1 application/user to another.
If I do it right the user won't even care or notice.

I appreciate people have mandates and I'm not saying don't make things
an ISO standard - for some people that is a selling point. But the
pursuit of a 'one size fits all' standard is a failed one. If you go
in with the attitude that '100% of our data must fit into this
format'. Well.. GML anyone? And oh by the way, I bet you'd still never
get true 100% adoption. There's never an exception.. until there is.

I go in with the approach that you'll try to tackle the majority
(maybe 80%) of your requirements with 1 or 2 solutions - and then mop
up the rest. Excessive future-proofing and chasing the tail will kill
things well before they come to fruition.

>
>> WMS has been a pretty good success even though I'm sure that'll get
>> some snickering from the peanut gallery due to its age - it is still a
>
> Ah - but a standard is good if it lasts, isn't it? Imagine having http
> change every half year. Wouldn't that be fun? There sure would be more
> work to do for us. HTTP 1.1 is 176 pages[1].

Oh by no means am I slighting WMS. I just thought someone might slight
us for still using it :) WMS is a great protocol. It's simple..
simple.. and it works! I can write a rudimentary client that grabs an
image from a WMS server in mere minutes. Granted dealing with an image
is much simpler than actual features.

The funny thing about WMS that I didn't mention - is that even for
such a short format - people don't even implement all of it! When's
the last time you saw someone using styles?

> You are right, GML sucks bad. But it is not true that it is not used.
> The German Cadastral and Surveying Authorities of the Länder adopted GML
> as core for the new cadastral base map format. They could only do so
> because it is an ISO de-jure standard and can only thus become part of a
> law. Technically it is a huge PITA. Just like cadastral bas maps are.
> But it also makes sure that folks now don't fall off the plate when they
> step over the border of their city boundaries, county or state. This is
> an achievement that neither proprietary vendors nor foss hackers managed
> in the whole cadastral IT history. The use case is just a bit different
> than locating the next pizza palace. The cadastre maps ownership - the
> basis of our whole economy (be it broken or not, this is what we live
> in, on and off).

Hence why I said 'barely' used. And here's the thing.. Realistically -
since GML is not widely adopted.. is the format 'really' open?
In other words.. if users can't access it then who cares? Maybe
putting it all in shapefiles would have been better!

(I'm not advocating that.. just making a point).

Now.. I do understand that sometimes you need an organisation to make
a policy decision to force adoption. I get that, I really do. But it's
not just an academic argument - if you have users that really need to
get the data.. well you can't just hand them the standards document
and tell them to write a client. Because again.. ultimately users
don't care about formats.. they care about content. They want
information they can use.. not data. Putting data in an inaccessible
format - open or proprietary - is ultimately no different than locking
the drives in a safe. If I can't use it it doesn't exist.

And of course that's the hard part.. getting from here (the
proprietary status quo) to there (the open world of unicorn
happiness).


> KML is more interesting from the governance perspective. And I am pretty
> happy that KML is not owned by Google any more but by the OGC because
> the OGC is a non-profit organization dedicated to make the world
> interoperate. When KML came out everybody was full of praise for the
> pragmatic way of doing things. Now that it is in the OGC it sucks again?
> Funny.

I'll honestly say that I've always disliked Google Earth from an
organisational perspective for one simple reason.

Mission Creep.

Non-GIS users look at it and see.. maps! With stuff on it! Hey that's
GIS right? The problem is from a enterprise standpoint is.. Google
Earth is just a viewer. And there have been a lot of attempts to
shoe-horn in GIS-like capabilities but ultimately it is a viewer that
is a silo where things go in but don't really come out. I realize KML
can exist apart from Google Earth, but we're not using it that way and
have no plans to.. I haven't seen anyone that does. Perhaps if GML
didn't exist then we'd start to see people using the KML syntax to do
GIS? Who knows, that could be interesting.

So in other words, while I have no issue with KML itself,
operationally I've only really seen it associated to GE (which I
dislike).

>
> Google now has to ask a diverse bunch of spatial experts, geo
> professionals and neo geographers if their changes to KML are worth
> pursuing. And OGC makes sure that even you have a say in the public
> comment period. Not bad, huh? But it gets even cooler. You could be one
> of those diverse spatial experts, geo professionals and neo geographers
> and join the process right from the start! If you think it sucks, then
> you can say so right away. Why wait until many people have invested lots
> of time and written large incomprehensible documents? You are wasting
> other people's time by complaining *afterwards*.

Sure, point the way.


> No offense taken please. Maybe I got carried away just a bit.

Not at all. I'm sure I offended some people but that's ok. I have no
problem voicing my opinion even if it's not popular. Saying GML blows
is probably heresy to some but it's the truth. CAP is another standard
I predict will fail. Good idea but just ultimately way too
complicated. Someone could probably do a thesis correlating schema
length to failure rate for XML and similar document formats.


 - bri

-- 
Brian Russo / (808) 271 4166