Agenda: 1 SLD vs Style 2 PostGIS Schema 3 SVN Down
jgarnett Meeting time!
jgarnett Hi guys, sent a ping to Brent, James etc...
jgarnett How is everyone?
MatthiasB Fine. Thanks.
jgarnett Hey justin sorry hacking did not work out last week, darn laptops hate me.
jdeolive no problem, what can you do
rgould hmm.. I think SVN might have just gone down again
jdeolive i managed to get some good hacking in
rgould anyone verify?
jgarnett Well I can buy you beer - thursday
jdeolive works for me
jgarnett I was going to do a quick agenda like thing and then open up the floor.
jgarnett 1. Stylish branch coming home
jgarnett 2. Schemas (PostGIS MM / DataStoreSQL questions & plan)
jgarnett 3. SVN down, and scheduled maintainance
jgarnett Let me get the sys admin online for item #3 ... um Corey do you want to talk about #1 while I fetch Paul ?
chorner right... we've been hacking away at styles, trying to get the geotools implementation closer to the SLD 1.0 spec, while also trying to keep with the GeoAPI
chorner Jody added StyleEvents too... but that's getting renamed apparently
jdeolive thats great, i have been working on gtxml binding for sld
chorner we're at an impasse at the moment, trying to decide if we should be firing the events on the StyledLayerDescriptor or Style
chorner StyledLayerDescriptor is rarely used in geotools at the moment – should we be using it or Style
jgarnett Well the spec says SLD but the code says Style? Sigh. Is this a mistake, and if so how do you guys want to fix?
rgould Paul is saying that cruisecontrol took down the server
jdeolive based on the maddness that goes on during the build, i am not surprised
rgould he should be here in a minute
jgarnett Hey Brent/Justin, how does GeoServer handle styles? It keeps SLD docs right?
jgarnett Brent - we are trying to figure out why Style is used everywhere in GeoTools, but the document starts with StyleLayerDescriptor (with a style in it).
jdeolive yup, you can sent it xml in a q request
jgarnett Let me try again, if we force the use of StyleLayerDescriptor throughout the code base (rather then Style) how messed up would GeoServer get...
jdeolive probably the same amount of messed up as udig
jgarnett Or phrased another way, how does GeoServer process the SLD, does it just take the First Style in the document?
jgarnett Well you guys do something right now, right now uDig does not do anything ...
jdeolive I didn't even know that a StyledLayerDescriptor class even existed
jgarnett It is the root node in the document, the parser creates it...
jgarnett James just went dark, so can't ask him right now...
jdeolive i would have to look into things a bit deeper before i answered
jdeolive but if we change it we change it, when geoserver goes to trunk lots of things will break
jgarnett That is kind of the case anyways, most of the Style/Filter interfaces were not set up to use Factory & FactoryFinder correctly
bowens off hand I don't know what will be affected
jgarnett Okay ... thinking
jgarnett I don't mind breaking code when it is something obviously wrong - like StyleFactory.create becomes StyleFactoryFinder.create
jgarnett but the StyleLayerDescriptor vs Style thing is a different sort of mistake ...
jgarnett It is a mistake in use, not in interface ...
jgarnett Okay this is dragging on ... I guess we will just do something (tm)
jgarnett Does anyone else have comments, or shall we move on?
chorner is having an SLD genuinely useful, or is it just a pretty package containing the Style (which we actually use)...
chorner Geoserver people? any thoughts?
jdeolive i guess it potentially could be, I mean there is stuff in the spec that the Style object does not capture
jgarnett Cool james is going to drop in and try and help...
jdeolive it could end up there perhaps
jdeolive but i dont relaly know the interface that well
jgarnett I think there is NamedStyle and UserStyle ...
bowens I couldn't say. But I think it is just the style that it cares about
jgarnett The idea that you set up a single document, and you use it for your holl "map", the renderer grabs each Style that matches when drawng a specific layer ...
jgarnett Hi James!
jmacgill Hey!
bowens I haven't looked at the two that much and I'm not sure of the differences
jgarnett I will quickly backtrack - we are stuck between Style and StyleLayerDescriptor
jgarnett right now geotools renderers take one, but the standard (and parser take the other).
jgarnett We are not really sure what GeoServer does with an SLD
jgarnett uDig just grabs the first style and makes use of it ...
jmacgill I'm not sure what GeoServer does, but the right answer is to grab all of them and store by name
jdeolive i am sure it does the same thing, they are using the same renderer
jgarnett This is an issue because I am considering trucking around StyleLayerDescriptor in uDig (w/ events) and would like some guidence on if SLD should creap out into the geotools API
jmacgill the higher levels of the SLD document have always been undersupported in GeoTools
jgarnett jdeolive, we do this before renderers enter the picture ...
jmacgill basicaly above FTS was never properly supported
jgarnett James on the stylish branch we documented and split out interfaces for the higherlevels...
jmacgill excelent
jgarnett Okay so we got once answer, process the document and store the Style objects by name. When it comes time to render we do multiple passes - once for each style?
jgarnett (in order of document appearance?)
Jesse_eichar we already hand FTS like this.
Jesse_eichar It is the multiple UserStyle objects that is the problem I think.
jmacgill for multiple styles, in WMS as they are 'named' then the query string tells you the order
Jesse_eichar ( I assume that STyle maps to User style )
jmacgill or even if all are used
chorner yes, Style == UserStyle
Jesse_eichar At the rendering level a layer has been mapped to a Style and that style is used. So I don't think the SLD/Style is and issue for rendering.
Jesse_eichar Except if a renderer wishes to register for event.
Jesse_eichar If events are only on the SLD then there needs to be a way to get the SLD.
Jesse_eichar Maybe the issue can be side stepped by adding events to style as well.
ph1l0r jmacgill may i interrupt you guys for a second. i'm trying to find a copy of your A* Java Applet. Any chance you can help me with a copy?
chorner the event (delta?) will tell you what the SLD is...
jgarnett ph110r - not sure what you mean?
chorner ...but you need the SLD in the first place to register the listener
Jesse_eichar One of the problems (I think) is that SLD is basically ignored everywhere. and now there are events on it so suddenly it is REALLY important. Is this a good idea?
ph1l0r http://www.ccg.leeds.ac.uk/james/aStar/
jgarnett chorner - we could move the events to Style? But does this mess us up in terms of supporting the actual document?
CIA-7 rgould 1.0.x * r17145 udig/features/ (4 files in 4 dirs): Forced source plug-ins to unpack themselves after build
chorner another note, is that Style applies to a layer, while SLD applies to a Map... so we're moving up a level...
jgarnett ph110r that is GeoTools 1.0, not sure it is activly developed.
jmacgill whats with the aStar reference?
jmacgill (nothing to do with geotools)
jgarnett james / chorner can you confirm that ?
ph1l0r jmacgill i'd love to have it
chorner good question... should be ok
ph1l0r and the link is dead
bowens can you give me some background on what these events will do?
jmacgill the server it is on died
jgarnett fire when the style is editied
jmacgill and I don't have an active machine with the source on
jgarnett (I would like to use the same events on Filter/Expression etc...)
ph1l0r then i'm out of luck i guess. all other java applets covering A* seem to be worse than yours (at least what i know from reading)
jgarnett (the internet never forgets does it James)
jmacgill it was on feathers
jmacgill which got rooted
ph1l0r his applet is linked from like a dozen sites
bowens oh, so when the user edits the style on the fly...
jgarnett yes, problem is doing the work revealed that we are not using SLD correctly/consistently in geotools.
jgarnett So the question for James is bascialy - how should we be using this stuff?
bowens gotcha
bowens what was teh incorrect use?
jgarnett Style
jmacgill yeah Syle should = NamedStyle or UserStyle
jgarnett Okay talking w/ Jesse - I think I got it straight.
Jesse_eichar (If I know what I'm talking about :-P)
jgarnett The MapContext should have the StyleLayerDescriptor, and should process it
jgarnett so by the time the renderer gets around to rendering a specific Layer it can ask the MapContext for the Style(s) it should draw with ...
jgarnett Is that consistent James?
jgarnett So import/export SLD needs to be assoicated with Map in uDig, not Layer?
jgarnett (makes me miss SymbologyEncoding documents - aka a single FeatureType style)
Jesse_eichar that's my understanding.
Jesse_eichar since SLD has UserLayers and UserStyles
Jesse_eichar Any opinions on what we should do with SLD? should we not worry about it and be able to add listeners to Style?
Jesse_eichar Should we make a pointer from Style to SLD so you can get the sld and add listeners to it?
Jesse_eichar should we replace all references to Style with SLD (rather not)
jgarnett Well we at least have a pointer up to the SLD, so option 2 is a go
jgarnett Option 1 bugs me but is easiest
jgarnett Not sure on option 3 - does not even seem to be the correct thing to do?
CIA-7 rgould * r17146 udig/features/ (4 files in 4 dirs): Removed unpack=false from the source features
Jesse_eichar I vote for the getSLD() method or something similar.
chorner getSLD only on Style?
jgarnett Well we have that ... I am concerned about option 3 - is it correct or not?
Jesse_eichar I don't think it is.
jgarnett getParent() will do .. someway to find the "root" so you can register for events...
jgarnett I would agree, do we have anything the processes a MapContext? Do we even care for GeoServer/uDig useage ...
jgarnett Okay we are stuck, and it is on an issue of Style/SLD use. I know enough to finish up the event system.
Jesse_eichar uDig doesn't care. It builds the MapContext manually because it has already made the mapping.
jgarnett I trust we can revisit when/if MapContext / GoObjects have a beat down.
Jesse_eichar At a different level, IE drag/drop or import we will care but not at the layer/renderer level I don't think.
Jesse_eichar I think MapContext should have a set SLD or something similar at some point myself.
jgarnett I think so as well.
jgarnett Okay we got enough that Cory and I can work ...
jgarnett moving on.
jgarnett 2. Schemas (PostGIS MM / DataStoreSQL questions & plan)
jgarnett jdeolive - this is all about you
jgarnett It seems we have two things a) bug b) issue
jgarnett a) bug, the "Schema" is not being included in the request for the spatial table
jgarnett (whatever that means)
jgarnett b) we lack a proper API to figure out which schemas are available
jgarnett uDig has a hack (for each and every JDBCDataStore) to handle the second thing. Since it is repeated it may be fun to promote this hack to an SQLDataStoreFactory interface
jgarnett (and this would give a home for some of the ConnectionPool methods GeoServer users have been asking for)
jgarnett Does this make sense?
jdeolive the postgis datastore contains api for setting the schema
jdeolive so its there, its probably that some queries are made without using that value
jgarnett yep - pretty sure it is just a bug.
jdeolive I agree
jdeolive it is on my list of things to fix for next geoserver release
jgarnett Does the second idea make sense? You have been following the GeoServer DataSource servelet configuration saga more then I ahve ...
CIA-7 magnasound * r17147 udig/community/lavila/plugins/org.cgiar.cip.diva.data.qualitycontrol.assigncoords/src/org/cgiar/cip/diva/data/qualitycontrol/assigncoords/ (5 files in 3 dirs):
jgarnett Okay do you have a Jira for it - or can I set one up for you?
jdeolive I think once the postgis datastore lives inside teh container the connection pool can be injected in
jdeolive the datastore factory already takes it as a paramter in teh constructor
jdeolive does that answer your question
jdeolive one sec, i will find the jira task
jgarnett Almost, w/ respect to setting up an SQLDataStoreFactory - do you care one way or another?
jgarnett What you said makes sense thou
jdeolive http://jira.codehaus.org/browse/GEOS-367
jgarnett Okay that is it for the agenda, seems Justing and Paul figured out what was killing svn ...
jgarnett Floor is open ...
jgarnett aside: did you guys have a meeting last week - did not see any logs