#geotools rschulz H ~email@example.com (Rueben Schulz)
#geotools cholmes H@ ~firstname.lastname@example.org (New Now Know How)
<rschulz> Merry Christmas Chris
<cholmes> Thanks! Merry Christmas to you too.
<rschulz> Looks like this will be an uneventful meeting
<cholmes> Yeah, especially considering you and I work on almost opposite poles of the geotools
<cholmes> Though soon I will be doing coordinate projection stuff in GeoServer, though I am a
+bit scared of it all...
<rschulz> Well, any plans to use cts...
<rschulz> You beat me to my thought
<cholmes> Yeah, I'm hoping to just use it in DataStores...
<rschulz> Hopefully the tutorial will help you, the concepts are not too bad on the user side of things
<cholmes> We need to get all the datastores returning their crs information...
<cholmes> and then queries that can request in a different crs and be able to do the
<rschulz> All users need to know about is CRS's and math transforms
<cholmes> And then I need to figure out how to make it all accessible to my users, like which
+projections they want to offer, and return them appropriately in a capabilities document.
<rschulz> I would assume WFS is just using epsg authority numebers (ie 4326) to refer to CRS's?
<rschulz> back to DataStores...
<cholmes> One thing I would eventually like to try to do is allow datastores to do the
+transformations themselves - like postgis can handle the stuff, using proj and its native
+data structures, so it seems better to just have it do it, instead of waiting till it gets
<cholmes> Yeah, basically just epsg numbers
<rschulz> shapefile should be easy to do properly once we start using the new wkt parser (and shake the
<rschulz> postgis has wkt in its spatial_ref table, so that should be easy to parse as well, to create
<cholmes> Yeah, that's what I figure - I just need some time to do it. Or procrastinate enough
+so it hits the udig guys Though that's kinda mean.
<rschulz> I do not know about the others and think that some of the data formats do not store crs info
+anywhere (ex. arcgrid)
<cholmes> Yeah, well I can have users input an epsg number for data that doesn't...
<cholmes> I have them do that now - it just doesn't do anything on the backend - returns it
<rschulz> Actually, if we get the wkt parser right, it should only be a matter of a few lines of code
+to add crs support to postgis (read only)
<cholmes> I suppose I'll have to do arcsde and oracle myself as well... Since I seem to be
+psuedo maintainer of both of those. And actually mysql as well
<cholmes> Though mysql is generally easy if postgis is done, since they both use the simple
+features for sql spec.
<rschulz> oracle uses some "form" of wkt if I remember correctly
<cholmes> Though I think the first thing I want to hit is raster support.
<cholmes> Have you used the grid coverage exchange stuff?
<cholmes> Jody was encouraging me to rewrite, though I really don't want to, mostly because I
+have no real GIS education and don't really understand all the theory behind them at all
<rschulz> no, it is still a mystery to me - I have been waiting for someone to write a good tutorial
<cholmes> I read the OGC coverage abstract spec over the weekend and was completely confused.
<cholmes> I've heard that the OGC does not do a good job with rasters in general, and then I
+think whoever did the grid coverage stuff didn't even implement that well.
<rschulz> martin does have some good javadoc notes
<cholmes> Perhaps I'll look at GML 3 and see if they have any good inspirations.
<cholmes> On grid coverage? Is he the one who wrote them?
<rschulz> grid coverages have been rather neglected
<rschulz> yes, part of his phd work involved remote sensing
<cholmes> Yeah, unfortunately they're my most popular feature request at the moment, so I may
+end up doing them.
<cholmes> And I even sorta told Jody I would try, since he needs them for udig.
<cholmes> But I feel way out of my element. I'll just have to spend a good bit of time doing
<rschulz> Please write anything you learn in a tutorial
<cholmes> Do you know how the old DataSource raster hack worked? I'm sorta contemplating
+something like that.
<cholmes> Will do - though I imagine if I do anything I'll probably just get frustrated and
+rewriting the classes when I figure out how I want them
<cholmes> So hopefully a tutorial would be less necessary, or would be like the same as
<rschulz> I think grid coverages also have a method to allow opperations, so there should already be
+code to do re-sampling (needed for crs changes)
<cholmes> That's what I really want, for users to just be able to input params, the same as for
<rschulz> the DataSource raster hack was replaced by the grid coverage exchange (which I briefly got
+involved in last august)
<cholmes> Was the hack super bad? Was the gce any better?
<rschulz> the hack was only to read rasters from a file
<cholmes> And we want gce to read wms too? Is that the deal?
<rschulz> hack was bad, gce is better but still has issues
<rschulz> yes (then you can get the wms data as a grid coverage)
<cholmes> Can you tell me what the issues are? Where it should be improved?
<cholmes> I may try to tackle it soon, though I have like 5 things I want to tackle in
+geotools, and I'm swamped this month. But my users would love it.
<rschulz> we wrote some of them up on the email list (search for jody's and my messages from sept)
<cholmes> Ok, will do.
<rschulz> the main issue was that we want DataStore and GCE to be similar: DataStore gives you features
+and GCE gives you grid coverages
<rschulz> unfortunately the concepts used in gce interfaces do not exactly line up with datastore
<cholmes> Yeah, I think what may be in order is to get coverages to implement feature, or at
+least to share a common super class...
<cholmes> Though I've not researched enough to really say that.
<cholmes> Which concepts?
<rschulz> I do not know how you would do that, since they are different beasts (though that was part of
+the DataSource hack, it returned one feature with the grid covearage as one attribute)
<cholmes> Well, there is a lot of theory about it all, and in GML 3 they do that, I'm pretty
<cholmes> But its very far up the hierarchy where it splits.
<cholmes> I'm not sure it's the right way to go, but it might be interesting to check out.
<rschulz> concept, GCE has a finder interface (which is like using a DataStore constructor) and a super
+fancy gce interface that will chose the proper finder base on an input url for a dataset
<rschulz> both return a GCEReader or Writer
<rschulz> This is similar to the ideas in the DataStoreFactory stuff, but different enough to confuse
<cholmes> Ok, I should really just dig into the code.
<cholmes> Are you having any trouble going to refractions.net ?
<cholmes> or is it just me?
<rschulz> yes, it is very slow
<rschulz> (but worked eventually)
<rschulz> some gce info (parts that I understood) are in the GCE tutuoral on the wiki
<rschulz> also remember that gce was to become part of geoapi, and the new grid coverage stuff will
+also implement geoapi interfaces (org.geotools.coverage)
<cholmes> But you never actually got the gce tutorial stuff working?
<rschulz> the new grid coverage interfaces are almost identical to the old stuff
<rschulz> I kind of got it working (got a raster displayed on a map), part of my problem was that the
+new gce implemetations had many bugs...
<cholmes> Which one were you using?
<cholmes> And do you have any sample files? Does our spearfish example have some?
<rschulz> I had trouble learning the new api, since I ofter encountered bugs when writing code that
+"should" have worked
<rschulz> the new spearfish and raster veiwer (name?) in the svn demo directory were updated to gce
<rschulz> I got it woring with arcgrid, image and I think geotiff
<rschulz> someone more knowledgeable than me needs to spend a few weeks with this stuff...
<rschulz> (it was suposed to be in geotools 2.1)
<cholmes> Cool. Well hopefully I can work through some basic bugs, and then just inflict it on
+my users, maybe get some bug fixes out of them
<rschulz> good luck!
<cholmes> Yeah, I think I'm going to need it.
<cholmes> And I don't know that I'm going to have much work time this month...
<cholmes> But WMS raster support just got another 2 votes in my poll:
<rschulz> yes, though I do not remember it being too bad once I figured out the interface
<cholmes> Cool. I suppose I will implement with it as is, and then after that see if I can
+redo it to make it better. It's good to hear that you actually did get things working though.
<rschulz> perhaps we can get jody to put some more time into this, udig is suposed to have raster
+support (hence all their time on wms)
<cholmes> Yeah, but I think I made a deal where I traded the time I promised on metadata (since
+I kept killing all their interfaces), to do raster stuff for him, since I actually care about
+it, whereas I don't care at all about metadata.
<cholmes> So you're in school now? Are you doing a masters or a phD?
<rschulz> masters of science (forestry), mostly statistics and modelling stuff, using spatial data as
<rschulz> I am still not making full use of the spatial side of the data
<rschulz> (University of British Columbia)
<cholmes> When do you graduate?
<rschulz> as soon as I complete 4 more credits and write a thesis (ideally by the end of the summer)
<rschulz> so I plan to do no geotools development over the next few month (will still do some work on
<cholmes> I've been contemplating going back to get a masters at some point...
<cholmes> But it's so much fun to procrastinate with geotools!
<rschulz> I am madly trying to finish of the crs stuff I am woring on before jan classes start
<rschulz> what subject would you like to study with a masters (geotools related?)
<rschulz> yes geotools is much funner, no hard deadlines...
<cholmes> Yeah, maybe GIS, maybe computer science, maybe something more applied...
<cholmes> Maybe something with open source.
<cholmes> I got a scholarship to study for 10 months in zambia, looking at 'the potential for
+open source gis software to help develop sdi's in developing countries' (roughly).
<rschulz> os, a rather broad subject for a masters degree
<cholmes> Which will be very vague, I'm not sure exactly what I'm doing, but it could be cool
+to spin that into a masters in some way.
<cholmes> True, probably would narrow it with both GIS and with developing countries.
<rschulz> cool, I need to start looking for work like that once I get out of here
<cholmes> Though yeah, there are many ways to look at it, I think it could be cool to do
+political science, and look at open source through that lens.
<rschulz> IanT's os gis book might be a good place to start
<rschulz> how so (poly sci)?
<cholmes> Yeah, you definitely should look for opportunities like it, there is a suprising
+amount of money available to do cool shit, you just have to ask for it, get creative about
<cholmes> Well, OS opens up a different way of organizing, and has lots of implications for
+civil society, doing things more transparently, letting order emerge out of chaos, ect.
<cholmes> My friend is doing a phD at berkeley, and one of her poli sci profs wrote a book
+about open source.
<rschulz> cool, I am constantly amazed by what "we" are able to acomplish in os
<cholmes> I actually haven't read that paper, but it's by the prof I was talking about.
<cholmes> Yeah, the more I do it the more convinced I am to never work for a commercial
+software company in my life.
<rschulz> thanks for the link
<cholmes> It's also interesting to think about its implications past software - the argument
+that software was the first because the people invovled were closest to the medium that makes
+it possible - the internet.
<cholmes> But that more and more professions can take advantage.
<rschulz> yes, I was at an esri course in oct and constantly surprised by the amout of time they spent
+discussing licensing (cost, how to do it properly, restrictions ...)
<cholmes> Very cool article on 'open source politics':
<rschulz> I have been eying up the wikipedia gis article for work, yet another step out of software
<cholmes> And my company's next project is hopefully going to try to work on some of the issues
<rschulz> bookmarked, thanks
<rschulz> which issues?
<cholmes> Well, what we're looking to do is to provide a sort of 'sourceforge for community
+groups', or for political activists, or really most anyone interested in open democracy type
<cholmes> Just as source forge encourages open source software projects, so we would encourage
+open democracy type projects.
<rschulz> so you would provide tools for these groups and hosting services?
<cholmes> Giving people web space, email lists, wikis, ect.
<cholmes> That they could use as appropriate.
<cholmes> Yeah, I think it would be a hosted service, where most anyone could sign up, and use
+the parts that they wanted.
<cholmes> So you wouldn't need to know a sys admin if you wanted a web page, or a wiki.
<rschulz> cool, that should prove very useful to many groups
<cholmes> And it wouldn't be a barrier to entry, that only more highly funded campaings could
<cholmes> Yeah, we shall see, that is the hope.
<cholmes> We just need to implement pretty well, make it super easy to use, and have it be
<rschulz> yes, sourceforge even confuses me, not a good implementation for non-computer people
<cholmes> Exactly. WHich is why we're not using their software.
<cholmes> We're looking at Zope/Plone...
<cholmes> Which is about the nicest content manager.
<rschulz> hopefully your project could go on to foster inovation on new colaboration software
<cholmes> Yeah, that's the hope, roll everything we develop back into the open source
<cholmes> There are thankfully some interesting starts for us to build on in the zope word.
<rschulz> I heard about it, a professor here was interested in it for organizing foresry info on the
+web (a portal for aquatic info was his inspiration)
<cholmes> Yeah, they've got an amazing community.
<cholmes> I'm hoping to integrate W*S with it as well.
<cholmes> Combine geoserver and this project.
<cholmes> So people would be easily able to make and annotate maps.
<rschulz> (project mentioned before www.onefish.org)
<rschulz> good luck, I have not seen this done well yet (but I have not had a chance to work with web
+gis yet, so have not seen much)
<cholmes> I work all the time with web gis and have not seen it done very well
<rschulz> I would love someone to set up a wikipedia for free gis data, but the quality control issues
<cholmes> The great thing about open source though is that when you come up with a good idea
+there's probably already someone working on it, that you can join with:
<cholmes> Yeah, it needs to be a bit more than a wikipedia.
<cholmes> I think the key is peer review.
<cholmes> Leverage the power of social filtering
<cholmes> Make it really easy for people to rate data, so good data sets rise to the top.
<cholmes> And also make it far easier to fill out metadata.
<cholmes> I actually read a great article on some solid ideas for sharing gis data.
<rschulz> a versioning system for spatial data would also be needed (and an automatic method to get
+metadata out of a users gps so they do not have to bother to do it themselves)
<rschulz> thanks, more reading for procrastination
<rschulz> well, I should get back to editing my report, nice chatting with you Chris
<cholmes> You too, have a good new year's eve.
<rschulz> ... I put up a copy of the log
<rschulz> Thanks, Happy new years too
<rschulz> no, should I ?
<cholmes> Oh, I don't think it matters either way.
<rschulz> well, I will stick one up for any interested in the current state of gce
<rschulz> good bye
<cholmes> See you later.
GeoTools : 4 December 27 2004
#geotools rschulz H ~email@example.com (Rueben Schulz)