IRC Logs, April 28th, 2008

0) what is up 1) paging 2) Hello to SoC H2 student

18 <aaime> Hi?
19 <aaime> zero topic meeting?
19 * jdeolive loves those kinds of meetings
20 <aaime> lol
20 *** jdeolive sets the channel topic to "Weekly IRC: 0) what is up 1) configuration".
20 <dwins> i've asked the student I'm mentoring to stop by and say hello this week, should I tell him to come back when people are around?
20 <dwins> (for gsoc)
20 <aaime> shouldn't we talk/vote about paging?
20 <aaime> groldan?
20 <jdeolive> oops
20 * jdeolive forgot what day it is
20 *** jdeolive sets the channel topic to "Weekly IRC: 0) what is up".
20 <groldan> hi, got distracted hacking paging
20 <aaime> jdeolive: ha ha
20 <groldan> yeah, add paging as topic please
21 *** aaime sets the channel topic to "Weekly IRC: 0) what is up 1) paging".
21 <aaime> anything else? we're already 20 minutes deep into the meeting hour
22 <aaime> acuster, groldan, jdeolive, jgarnett
22 <aaime> anything else?
22 <groldan> not here
22 <jdeolive> not sure if jody is around today... he may be stuck moving
22 <aaime> aah right
22 --> simboss has joined this channel (n=chatzill@host201-203-dynamic.27-79-r.retail.telecomitalia.it).
23 --> wohnout has joined this channel (n=wohnout@kolej-mk-60.zcu.cz).
24 <dwins> aaime: can you add 'say hello to the SoC student working on H2' as a topic as well?
25 *** aaime sets the channel topic to "Weekly IRC: 0) what is up 1) paging 2) Hello to SoC H2 student".
26 <aaime> who's running the meeting?
26 <groldan> I thought you
26 <aaime> eh, I knew I would end up doing it
26 <aaime> Ok
26 <aaime> 0) What's up
27 * aaime flying low by at a high speed with DuckHawk project
27 * groldan hacking on paging
28 <simboss> doing nothing
28 <acuster> acuster — writing a file to clean the SVN dump
28 * jdeolive is working on rewriting component wms spec
29 * groldan does not understand what jdeolive is doing
29 <jdeolive> the spec as written is confusing and ambigious... so ogc is getting topp to make it so that it is not
30 <groldan> cool, thanks for the clarification
30 <jdeolive> np
30 <groldan> next topic?
30 <aaime> 1) paging
31 <groldan> okay, I was hoping to ask for a vote
31 <groldan> and was going to test postgis but the meeting time catched me up
31 *** aaime is now known as gt-meetbot.
31 <jdeolive> HAHA
31 <groldan> yet, I'm leaving QueryCapabilities.getFilterCapabilities() out of the picture by now, since its being an unexpected scope increase
32 <gt-meetbot> groldan, it was something requested by Jody as a condition to let the proposal go afaik
32 <groldan> do we have enough PMC around for a vote? guess not, but does anybody read the proposal?
32 <gt-meetbot> I wrote most of it... does it count as reading?
32 <groldan> thanks gt-meetbot, yeah, it was about having a default implementation with everything set to false
33 <simboss> groldan can I ask a quick question?
33 <groldan> sure
33 <simboss> do you mention
33 *** gt-meetbot is now known as gt-clown.
33 <groldan> gt-meetbot: that certainly count, I wish I have more bots that write them
33 <simboss> I have playing a little bit lately
33 <simboss> with ebXML registries
33 <simboss> where paging is part of the spec
33 <groldan> having a lot of xml fun so
34 <simboss> (using JAXR )
34 <jdeolive> i am weary of introducing new api...
34 <jdeolive> it is indeed scope increase
34 <jdeolive> not saying its a bad idea
34 <groldan> so you have some recommendation about paging?
34 <jdeolive> but it might be good to seperate out
34 <simboss> paging there is implemented using offset and maxelement
34 <groldan> how thar relates to CSW 2.0 paging? the same?
34 <simboss> I quickly scanned through the proposal
34 <simboss> of feature paging
35 <simboss> and there was no mention of maxelements/maxresult/numofelementshit or whatever you want to call it
35 <gt-clown> zzzzz
35 <groldan> maxFeatures?
35 <gt-clown> that was the idea
36 <gt-clown> keep on using maxFeatures, it's already there
36 <simboss> k then I scanned it too quickly
36 <simboss> thx
36 <groldan> we're reusing Query.maxFeatures for that concept
36 <simboss> question answered
36 <gt-clown> it may well be non mentioned at all (it's already in the query api)
36 <groldan> and I'm more about to call offset startIndex
36 <groldan> and let offset/limit as jdbc jargon
36 <simboss> startIndex is better IMHO
37 <gt-clown> maybe startFeature
37 <gt-clown> startFeatureIndex
37 <gt-clown> (to better match maxFeatures?)
37 <gt-clown> (offset/limit --> sql jargon)
37 <jdeolive> gt-clown: did we figure out what wfs 2.0 calls it?
37 <groldan> may be, though everybody seemed to worry about having a name reflecting some existing spec
37 *** gt-clown is now known as aaime.
38 <aaime> They don't call it at all
38 <aaime> results are treated like a linked list of pages
38 * jdeolive is going to start calling andrea "changling"
38 <aaime> Odo please
38 <jdeolive> haha
38 <aaime> you specify maxFeatures, the feature collection returned has a link to the next page
38 <jdeolive> i see
38 <jdeolive> alrighty
38 <groldan> yeah, they don't use it, just assume you start at the fist "page" and then follow server generated "links"
38 <groldan> no random access to pages
39 <jdeolive> i see
39 <aaime> (we can always add that as our custom vendor option for GET requests, people would love us if we do)
40 <groldan> and wfs 5.0 will include it
40 <jdeolive> anyways, to sum up, i am +1, no real preference on startIndex vs startFeatureIndex, but -1 on QueryCapabilities for now
40 <jdeolive> I think it should have its own proposal
40 <groldan> what do you do to know if the sorting will work?
41 <groldan> just catch up an exception?
41 <jdeolive> hope and pray
41 <jdeolive> the same thing we do with everything else in teh datastore api
41 <jdeolive> dont get me wrong
41 <jdeolive> i like the idea of QueryCaps... just think its a larger issue
41 <jdeolive> and dont want it to hold up paging
41 <groldan> it is if we include FilterCaps, which I'm leaving out so far
41 <aaime> remember it was not there to start with
41 <aaime> Jody asked for it
41 <aaime> in order to let the proposal get a +1
43 <jdeolive> right, and what i am saying is we should tell Jody to start a seperate proposal
43 <aaime> (in my first version of the proposal caps were not there)
43 <groldan> okay, in the mean time I'll make sure I get an end to end test ready
43 <aaime> sure, asking does not cost much
43 <groldan> and since we're in trunk, we can go with a minimal caps (ie, without filter)
44 <aaime> but given he warnted us it's probably better to ask before starting to commit
44 <groldan> and let depuring QueryCaps be another proposal
44 --> gdavis_ has joined this channel (n=gdavis@mail.refractions.net).
44 <groldan> jdeolive: would that work for you?
44 <jgarnett> hello - I am with a customer today (so am missing most of this meeting)
44 <jdeolive> groldan: yup, works for me
44 <groldan> hi Jody
44 <groldan> quick recap for you
45 <groldan> talking about paging
45 <groldan> we want to get minimal with QueryCapabilities
45 <groldan> ie, FilterCapabilities is being too much an scope increase
45 <groldan> but as we're on trunk we could start minimal
45 <groldan> and let QueryCaps mature as another propoasl
45 <groldan> sal
46 <jgarnett> bleck
46 * jdeolive misunderstood, his suggestion was to forgot QueryCaps all together for now
46 <groldan> that gives jdeolive's +1, otherwise he's -1
46 <jgarnett> I understand that QueryCaps is a larger scope
46 <jgarnett> however I would like to see either that; or someone taking the time to fix the existing datastores.
46 <jdeolive> if people really want QueryCaps right now i could be persuaded... but it seems unnecessary to me at the moment
46 <jgarnett> (ie I don't want to see datastores continue to suck; I was offering QueryCaps as a compromise)
47 <groldan> jdeolive: the gain I see is we won't need to change code later even if QueryCaps gets more methods
47 <jgarnett> I would be happier with doing the work to all the datastores; ...
47 <jdeolive> jgarnett: its a big changes... involves changing datastore implementations and client code
47 <jdeolive> it should not be taken lightly
48 <jgarnett> understood; so what change costs less?
48 <jdeolive> groldan: fair enough... so what does a minimal QueryCaps cover?
48 <jdeolive> just paging?
48 <jgarnett> FilterCaps? or fixing the DataStores?
48 <jgarnett> what I don't want is "half of paging"
49 <jgarnett> (ie the alternative presented earlier, Query2 with a FeatureSource2 is the least impact way of supporting Paging, ie new functionality = new method, new interface just like with GetObjectById)
49 <groldan> jdeolive: a minimal QueryCaps to cover isOffsetSupported() and supportsSorting(SortBy[])
49 <jgarnett> jdeolive; how would you do "just paging"?
49 <jdeolive> jgarnett: sorry, but its a reality of our datastore api at the moment, and its been worked around, you start chagning impelemtnations, work arounds stop working
50 <jdeolive> jgarnett: ok, i could go for that
50 <jdeolive> sorry, i mean groldan: i can go for that
51 <jgarnett> jdeolive; I got a few minuets to talk - we went over this on a couple of geoserver meetings. I don't mind what is done so long as it is complete - I am sick of half done work (like this StyleVisitor changes).
51 <jgarnett> groldan++ I can go for that too.
51 <groldan> cool, that's what I can too with the given timeframe
51 <jdeolive> jgarnett: i know, and i agree
51 <jdeolive> but drastic changes done work
51 <jdeolive> dont work
51 <jgarnett> yep.
51 <jdeolive> they lead to headaches
51 <jdeolive> its a big job... that is all i am saying
52 <jdeolive> putting it as a blocker to doing paging is a bad idea imho
52 <jdeolive> i am bnot saying it should not be done
52 <jgarnett> so what did we wend up with here; GetCaps with a single method isOffsetSupported() ?
52 <jgarnett> (looking for the proposal link...)
52 <groldan> paging implies sorting
53 <groldan> http://docs.codehaus.org/display/GEOTOOLS/Feature+paging+and+Query+capabilities
53 <groldan> so I mean isOffsetSupported(); and boolean supportsSorting(SortBy[]);
53 <jgarnett> makes sense.
53 <jgarnett> groldan are you going to update this page now? and call a vote during the meeting .... or are we expecting an email.
54 <groldan> nope, rigth away
54 <aaime> +1 for paging + limited caps
55 <jdeolive> +1
55 <jgarnett> +1 (you are updating the page now then?)
57 <groldan> done
57 <groldan> gonna add the votes now, thanks all
57 <jgarnett> QueryCapabilities as an abstract class please?
57 <simboss> +0
58 <jgarnett> for even as a final class; it is only a datastructure with fixed intent...
58 <groldan> not an interface
59 <jgarnett> sorry groldan; I am not sure when FilterCaps turned into an interface (it is in the proposal page)
59 <groldan> hmmm filtercaps is an interface afaik
59 <groldan> and I understood you wanted the same trick than for Query, ResourceInfo, etc
00 <jgarnett> I thought it was going to be a final class, or at least an abstract class - so that we can add more methods to it over time.
00 <jgarnett> (and not break everyone)
00 <aaime> he did want to avoid the same mistake as Query afaik
00 <groldan> yet, I was wondering if there's overlap in intent between ResourceInfo and QueryCaps
00 <jgarnett> not so much
00 <jgarnett> resourceInfo is about the data
01 <groldan> I see
01 <jgarnett> QueryCaps is about the FeatureSource API and how much you can trust it (ie interacting with the data)
01 <jgarnett> we got only a few more minuets; can we move on to the next agenda topic...
01 <jgarnett> or do you need more groldan?
02 <groldan> I'm gonna change to QueryCaps as an abstract class with no setters
02 <groldan> if that's what you meant
02 <jgarnett> 2) Hello to SoC H2 student
03 <wohnout> Hello that's me
03 <jgarnett> welcome!
03 * dwins lets wohnout do the talking
03 <wohnout> I have project to make spatial index to H2 database
04 <groldan> cool, welcome, btw
04 <wohnout> It should be great if it will be done
04 <wohnout> and I hope that I will make it
04 <groldan> so do you have a plan?
05 <aaime> wohnout, did you notice people were talking about your project in the H2 mailing list?
06 <wohnout> aaime: yes. I write there some info about me
06 * dwins was conveniently cc'ed on that thread
06 <wohnout> groldan: not so detailed but I have some ideas. I was talking to dwins and jdeolive today.
06 <jgarnett> I have a silly question; where is this work going to take place? As a geotools unsupported module? (and if so can i ask that we bring this one up to supported status before the end of the summer)
07 <aaime> jgarnett, I think we should've stated that as a condition before the student accepted
07 <aaime> raising the bar after the fact does not seem fair
09 <groldan> but what's the intent, being in geotools code base or another place?
09 <jgarnett> that was more my question, where is the work going to live.
09 <groldan> and if the former, we should help him as much as possible to, even if not getting to supported status, as close as possible
10 <groldan> (ie, by providing code reviews etc)
10 <-- DruidSmith has left this server (Read error: 113 (No route to host)).
11 <aaime> wohnout, do you understand what we're talking about?
11 <wohnout> I'm new to geotools ocde so I don't know how to answer
11 * dwins gets the impression that summer of code projects often have their own project on code.google.com
11 <wohnout> *code
11 <aaime> dwins, if they are stand alone, yes
12 <aaime> if this one heavily depends on gt2 api it's going to be easier to have it as an unsupported module in gt2
12 <-- gdavis_ has left this server (Remote closed the connection).
13 <dwins> sure
13 <dwins> but shouldn't gt2 use the database rather than the other way around?
13 <wohnout> I was looking on H2 module in gt2 and I have feeling that is not maintained by anybody
13 <jgarnett> well met wohnout; I gota run but it has been great to "meet" you online.
14 <jgarnett> jdeolive maintains the H2 module; but has not brought it over to be part of the standard "plugins" for geotools.
14 <wohnout> same to you
14 <aaime> wohnout, jdeolive here is the maintainer of that module
14 <aaime> wohnout, so the plan is to work on h2 codebase to provide a better spatial index
14 <aaime> and not to try and leverage the one alrady there?
15 <dwins> we have been discussing how the spatial index in h2 doesn't work well for non-point data
15 <aaime> if so, can I suggest a good book?
15 <aaime> "spatial databases" by Rigaux, Scholl, Voisard
15 <groldan> indeed
15 <aaime> has a few chapters on spatial indexes
16 <groldan> (though I may have read just 1/6 of it)
16 <wohnout> I think that this should be something above h2...
16 <aaime> groldan, me too
16 <groldan> but you had too much to do to finish it, I was not understanding
17 <aaime> well, indexes usually are integrated with the db for good reasons
17 <aaime> thought of course integrating requires more work
17 <aaime> ah, sqlite rencetly got a spatial extension
17 <aaime> you may want to have a look at it
18 <aaime> http://www.gaia-gis.it/spatialite/
18 <wohnout> ok will take a look on it
19 <aaime> (not sure it has a spatial index, but people were speaking highly of it at a recent conference)
20 <groldan> so that's it? go with our best wishes wohnout.
20 <groldan> should we call for a wrap, or do you want to say anything else?
21 <wohnout> thanks
21 <wohnout> no that's all
21 <groldan> okay, welcome aboard again
21 <groldan> bye all, bed time here
22 <aaime> Who's posting the logs?
23 <groldan> okay, that can be my good action of the day, I'll post them
23 * aaime loves groldan