Skip to end of metadata
Go to start of metadata

0) What is up?
1) Geotools 3
2) Dealing with feature validation
3) Governance
4) Graduation
5) Distributed Versioning Systems

aaime: 0) What's up?
desruisseaux: Martin: cleaning geoapi (doing metadata right now, fixing javadoc and removing a few deprecated methods).
***aaime is working on per layer security in GeoServer and the many funny ways in which people might need the server react to a non authenticated access attempt
desruisseaux: (posted an email on geoapi-devel about the deprecated methods to be removed)
***Eclesia working on styles and renderer
acuster: acuster: cleaning geotools, preparing math issues for isogeometry
***afabiani almost finished wcs 1.0 EMF bindings module
acuster: argh
***cbriancon working on coverageio / coverageio-netcdf to handle the GML in Jpeg2000 standard more properly
acuster ha scelto come argomento: Weekly IRC: 7 July 2008 — 0) What is up? 1) Geotools 3 2) Dealing with feature validation 3) why does no one show up to IRC meetings any more?
aaime: Next topic?
desruisseaux: GeoTools 3
aaime: 1) GeoTools 3
desruisseaux: Posted a proposal here:
desruisseaux: (just a proposal about the infrastructure, not yet the overall design)
jalmeida: my knowledge don't is to geotools developer , sorry
jalmeida: is for geotools user
desruisseaux: Main points:
acuster: wern't you planning to split out gt-util?
desruisseaux: * May take a year or two before it would become interresting to most current users
desruisseaux: Adrian, yes
phoster n=phoster@ è entrato nella stanza.
desruisseaux: This split could be done on GT2 before.
desruisseaux: * Would try to get Maven to compile with perfectly clean javadoc right from the begining, with attempt to keep it clear during all development process.
phoster: its possible create a filter that compares geometries ?
desruisseaux: Right now, I would not have to courage to fix GT2 javadoc (thousands of warnings...)
desruisseaux: I'm not a specialist of filter; we need to ask them.
aaime: (when the developer meeting is done)
aaime: desriusseaux, do you know when jee6 will be released?
desruisseaux: Would try to have only one feature model, etc. from the begining.
desruisseaux: No I don't know when JEE6 will be released.
desruisseaux: However it is going to be the first opensource Sun JEE.
desruisseaux: So it may help its adoption.
aaime: Hmm... unlikely?
aaime: The main reason why people are not eager to upgrade to new java versions in the server space
aaime: is the upgrade price of the commercial containers
aaime: (besides the fact that IBM is always quite a bit late in releasing a websphere that runs on the newer jdk)
desruisseaux: I think that we will have a least one year of work ahead before GT3 become usable. Maybe it would be long enough. But in the advant the it would JEE6 would still not ready, there is an alternative.
aaime: anyways, some articles around in the net seem to suggest jee6 will see the light this year
acuster ha scelto come argomento: Weekly IRC: 7 July 2008 — 0) What is up? 1) Geotools 3 2) Dealing with feature validation 3) why does no one show up to IRC meetings any more? 4) Graduation
CIA-34: acuster * r30938 /trunk/modules/unsupported/geometry/src/site/apt/review.apt: Cleanup: geom data are LGPL from JTS
aaime: so two years might be ok
desruisseaux: Sound a likely time frame.
aaime: Quesiton, in this two years I expect to see many advancements in the existing gt2 code
aaime: like real complex features and the like
aaime: but with no driver to move them to gt3
desruisseaux: We would not be ready to move them to GT3 before a while anyway, so it may let them the time to land in GT2.
aaime: so, gt3 is likely to be sitting without any vector code for two years?
desruisseaux: The blocker issue in GT3 (as I see it) is an ISO geometry module.
desruisseaux: Yes, it may take a while before we get vector code if we try to get geometry first.
aaime: like, I don't see any GeoServer related developer being interested in working on gt3 as is before we can use jdk6 for gs as well
aaime: and probably uDig too, since they claim they cannnot use 2.5.x either due to misbehaviour in the datastore events (gt3 will lack much much more)
aaime: (at least for a sizeable amount of time)
jalmeida: but geoserver users java 1.5
desruisseaux: GT3 would stick to metadata, referencing and geometry for a long time because of the time needed to develop the geometry part. When we will begin to port the other parts of GT2, there is good chances that they will be more advanced.
desruisseaux: Furthermore if we agree to use a distributed versionning system, it may be easier to maintains a JDK 5 branch.
desruisseaux: (because of the more advanced merging mechanism)
aaime: Ok, say that in these two yeras there is a driver to actually go full blown with complex features
aaime: and other api breaking changes
aaime: that would qualify for a GT3 name alone, no?
jalmeida: no is possible working with geoserver and openlayers,
jalmeida: or create one geo feature in java script for working with this tecnology
acuster: jalmeida, ?
desruisseaux: GT3 was just a temptative name; it we need something else in order to make room for complex features, we can find an other name.
aaime: jalmeida, that's a question for #geoserver or #openlayers channel I guess
aaime: Ok. I know this seems a little silly, but it seems like stamping that new effort GT3 might imply no major changes in the 2.x.y series
aaime: anyways, let's see what Jody plans for the future
aaime: the biggest change I see on the radar is real complex feature stuff but it does not depend on me so that might be just a red herring
acuster ha scelto come argomento: Weekly IRC: 7 July 2008 — 0) What is up? 1) Geotools 3 2) Dealing with feature validation 3) Governance 4) Graduation
desruisseaux: Okay, we can put the GT3 discussion on a rest for today.
acuster: looks like there's richer styles coming down the line too
jalmeida: i talk aboult geotools facilities
aaime: 25 minutes in the meeting, shall we move to the next topic?
jalmeida: to work with others tecnologies
desruisseaux: Fine for me, lets move on to topic 2.
aaime: 2) Dealing with feature validation
aaime: Well this is mine, and I've already written a mail on this topic on the ml
aaime: was just wondering if anybody read it and has an opinion about it
acuster: jalmeida, I have a hard time understanding you and how what you say relates to the topic we are discussing. Perhaps you could write longer sentences?
Eclesia ha abbandonato la stanza.
aaime: (was kind of hoping Jody to be around...)
aaime: If no one wants to discuss it we can just jump to the next topic?
aaime: 3) Governance
aaime: desruisseaux? acuster?
desruisseaux: Adrian I guess
acuster: So it seems like there is less and less participation in IRC meetings
acuster: PMC is no longer sticking to its mandate
acuster: and generally it feels like no one is taking things particularly seriously
acuster: The level of participation in the Copyright cleanup was risible
aaime: Mumble, about the missing people
aaime: jdeolive and simboss are in vacation
acuster: yeah, they are supposed to send messages out
aaime: correct
acuster: and it used to be they were supposed to read and comment on the IRC logs
acuster: no one has done that in living memory
aaime: but Ian Turton has not been around for ages in fact
acuster: and beyond the PMC are there any active module maintainers anymore?
aaime: Jody just got back from Poland, not sure he'll be around today
acuster: it seems not
aaime: I do maintain my unsupported module, which is kind of contractory (wink)
acuster: yes, there are reasons for everything, the question is whether the PMC system is still viable
acuster: anyhow moving on
acuster: 4) Graduation
aaime: acuster, wait (smile)
acuster: ok
aaime: do you have any suggestion on how to improve things?
acuster: not today
desruisseaux: Revisit the PMC member maybe?
acuster: I'm trying to make people aware there is a problem with GeoTools
acuster: so people can start to think of answers
desruisseaux: (I means the list of members)
acuster: seems the problem is more fundamental
oterral123 ha abbandonato la stanza.
acuster: the 'supported' modules are supposed to be supported, right?
aaime: eh yeah
acuster: where are those people? Why don't we hear from those maintainers?
aaime: but you're right that many supported modules are supported only as long as either uDig or GeoServer need fixes for them
acuster: seems like we should be pruning things down to a core set of modules which actually are active
acuster: even within geotools2
aaime: like I know I can open jira issues, but then I also know I have to fix them myself
acuster: sounds like unsupported to me
aaime: eh, the real unsupported modules are totally dead by comparison
desruisseaux: Well, actually back to the GeoTools 3 proposal, may mean motivation is not to bring a brand new API (I would try to keep it close to GT2), but rather to have a cleaner project to manage, exactly.
acuster: and as far as anyone pitching in beyond the 'what I need stage' I see precious little effort in that direction
aaime: but acuster, you're right, most people care about gt2 only as a reflection of the needs of the projects they are working on
acuster: personally, I'm not going to put any effort into geotools for a while
aaime: too bad, but thanks for the effort put in the graduation
acuster: can we move on?
aaime: sure
acuster: 4) Graduation
acuster: we missed this month's OSGeo governance meeting
acuster: which will happen friday the 11th
acuster: the next one might happen as early as August 1st
acuster: so anyone who cares can try to finish the provenance review of the rest of the unsupported module
acuster: if no one cares, we should probably delete the code to that module
acuster: s/that module/those modules
acuster: Cameron will be working this week/weekend at his initial final review
acuster: that's all I can think of
EdIPS è entrato nella stanza.
aaime: Ok...
desruisseaux: Do you means there is not enough time left before this week meeting?
acuster: no Cameron has work to do and he won't finish that before next week
aaime: this graduation thing is kind of silly because all the effort is basically on geomatys people + some Jody time
acuster: why "silly"?
desruisseaux: Okay, thanks Adrian for the information.
aaime: but at the same time, it seems they are the only one that do really want the OSGEO graduation?
acuster: this was a primary goal of Geotools 2 years ago
acuster: one that I didn't really care about but that has since taken up a lot of my time
acuster: the result is a good thing, the files are cleaner than they were
aaime: Agreed, it's an improvement
acuster: the geosolutions folk are cleaning up some of their mess
acuster: other ambiguous data is getting cleaned up
acuster: there are now formal copyright assignement docs on file
acuster: on the other hand, indeed no one apart from Jody, pitched in any real effort
desruisseaux: I have the very strong wish to make GeoTools a little bit less like a bazar and a little bit more like a cathedral... The legal issue was part of this wish; the other part being the API itself.
Eclesia è entrato nella stanza.
acuster: mixed metaphor martin strikes again
acuster: the essay was about modes of production not about the final product
acuster: but it's cute
desruisseaux: Thanks (smile)
acuster: aaime, what is your objection to distributed versionning systems?
acuster: which seems like it would solve a lot of issues
aaime: various, like ignorance about them and lack of interest in learning them
aaime: or the fact that development gets more and more disconnected
desruisseaux: I was reticent too, but the JAXB experience gave me the feeling that this kind of issues could have been avoided more easily with a DVS.
aaime: everybody doing his own thing in his own branch
desruisseaux: I realize that communication was probably in cause (which DVS would not solve), but DVS as a technical tool could have make things a little bit easier.
desruisseaux: This affair contributed a lot to make me change my mind.
aaime: the gt2 community is already going on parallel avenues, the centralized svn being the thing that keep us on the same page
acuster: I'm not sure I believe that argument
acuster: we already have folks doing duplicate code left and right
aaime: I am concerned that if we start using dvs we'll just foster a fork more
desruisseaux: This is an issue I have too, but I'm uncertain if it would happen.
acuster: I suspect someone would gain the 'authoritative' tree because it was cleaner, more complete, or some other criterion
acuster: or there could be an agreement to commit clean, finished code to the 'authoritative' server
acuster: and have an experimental server nearby (for all the unsupported stuff)
desruisseaux: It would reduce the pressure to commit stuff before they are ready or accepted.
aaime: I foresee a situation in which people jsut develop the code they do need, it's not mergeable back, and thus the forks do start
acuster: perhaps
pramsey ha abbandonato la stanza (quit: ).
aaime: also, people will somehow start developing code without making a proposal because they can
aaime: and then the proposal comes, but they have no driver whasoever to really merge back
KevinIPS è entrato nella stanza.
acuster ha scelto come argomento: Weekly IRC: 7 July 2008 — 0) What is up? 1) Geotools 3 2) Dealing with feature validation 3) Governance 4) Graduation 5) Distributed Versioning Systems
arneke è entrato nella stanza.
aaime: so any change that could have been discussed on the paper
aaime: and implemented is just discarded because the implentation is already there
acuster: what's a "driver" that would lead to code coming "back"?
acuster: some incentive?
aaime: given that most of the code we develop today has commercial backing, I'd say, a contract that asks you to merge back
aaime: I hardly see any motivation other that that
jalmeida: Graduation
acuster: well no one does that, nor writes javadocs, nor writes Jody's user docs, nor ...
aaime: acuster, I know we're working in a less than ideal way
jalmeida: my view is to make the best known is geotools
desruisseaux: Well, actually this kind of incencitive is the one I'm worrying about, because it lead to a merge "when the client wants it" which is often not "when the code is ready".
acuster: proposals were only ment for changes to the central core of geotools
jalmeida: one magazine
aaime: but I fear that with a dvs we'll get even worse
aaime: desriusseaux, you're assuming people will keep on developing on the code until it's ready
aaime: I don't believe that
desruisseaux: It depends on the context where peoples work
aaime: correct
desruisseaux: On our side, we are more on a scientific context
desruisseaux: I tend to put a lot of energy until I'm happy with my code, often much longer I was supposed to do
aaime: ok, so this reiterate my impression that gt3 will become a geomatys-only land
desruisseaux: Which explain partially why I'm always late
aaime: If I take consistenlty longer than expected I just get fired (wink)
desruisseaux: A little bit, but at this time its look like that geomatys is the group targeting the farest horizon
aaime: desriusseaux, do you see any other groups heading towards that direction anytime soon?
acuster: well is there any module that is being seriously improved by someone who is not the original author?
vheurteaux: aaime: desruisseaux is one of the owner of geomatys so it's difficult to fire him (wink)
acuster: seems like everyone is writing their own code anyhow
acuster: the fact that the javadocs for complex-feature are incomprehensible, even in geoapi, is testimony I think
aaime: vherteaux, ok, you do realize that is a very peculiar situation right? (wink)
acuster: the poor user who just showed up trying to understand things was visible discouraged
acuster: visibly
aaime: so you say, better no work at all than so so work?
acuster: if geotools is a library, yes
acuster: better to tell users we can't do things
acuster: than to make them take weeks to study our work
acuster: to find out we can't do them
acuster: or to give up in complete confusion
desruisseaux: Andrea, this is true that there is not many group publicly involved in GeoTools with a far horizon. This is exactly the issue we have with GT2 - a library developped on a "2 weeks deadline" basis doesn't meet our idea of a librarty created as a fundation for a long-term project.
jalmeida: if you want to integrate, cientify and comerciality, integrate geotools and you application into web developement
acuster: jalmeida, please, if you want to contribute to the discussion welcome, but save your other comments for later
aaime: I kind of disagree, I prefer to see code that can be improved by someone else than no code at all
aaime: if nothing, because that's the only code I can contribute to geotools
afabiani: sorry guys but I agree with aaime here
desruisseaux: This is a legitimated point of view, but I would like this code to sit in some kind of sandbox before they get merged in the clean trunk.
jalmeida: ok , sorry
acuster: but then you are telling your users you are willing to waste significant amounts of their time
acuster: that would be okay if it were an addition to the library which is why unsupported was a good idea
afabiani: gt2 has a lot of incomplete code and id someone wnats to use something working often should wait ehmm some years !!!!
aaime: acuster, I'm telling users I'm doing a gift of the code to them, if they don't like it I'm not forcing them to use it
acuster: but not as the core
acuster: yeah, that's not how free software libraries have traditionally been
aaime: acuster, if you see my commits on gt2 you'll see that I do constantly contribute to modules I'm not maintainer of
acuster: it's not "hey this is my junk that works for me"
desruisseaux: Adrea thats fine, but I would like to make a separation between code that we think will last, and code that live in a sandbox for now.
acuster: but "hey, we're trying to build a foundation that can wrok for lots of us"
acuster: look at all the C libraries
acuster: and you see a much higher level of code
aaime: acuster, this is all nice and dandy, but I don't have such a mandate, you'd have to discuss with cholmes about such plans
desruisseaux: The "unsupported" moduels was suppsoed to be a sandbox for the supported modules, but in practice the whole GT2 become a kind of set of unsupported modules.
desruisseaux: This is why I would like to create GT3; not because it would have brand new API, but because it would contains as clean as we can API, and I would use GT2 as the sandbox for GT3.
acuster: the problem with geotools 3 is that it's really geotools 0.3
acuster: but the "3" number makes it seem like a very mature project
desruisseaux: Lets try to make GeoTools 1.0 then.
desruisseaux: So in Adrian words, this is really that: I would really, really, really love to get out of GeoTools 0.2 and start creating (at last) GeoTools 1.0.
acuster: you think if we release another one of those, no one will notice (smile)
aaime: acuster, I agree with you in principle, but cannot participate in the effort in practice (just like the osgeo graduation stuff)
desruisseaux: Thats fine
acuster: it's too bad because at the top and bottom of the stack there is some great code: complex-feature and referencing
acuster: aaime, how much of your time is paid for as PMC work
acuster: 0 weeks a year?
desruisseaux: The first modules would be metadata, referencing and geometry, which are mostly geomatys stuff for now
aaime: acuster, correct, my participation is gt2 is justified only by the needs of geoserver
acuster: because that is a problem
acuster: Justin is with you all as well, right?
aaime: yes
acuster: so there are 2 PMC members with no mandate to do their PMC work
aaime: we have our mandate to do PMC work
aaime: it is to make sure gt2 will keep on working for GeoServer, and to try hard to contribute back to the gt2 project
acuster: how is that PMC work?
acuster: PMC members:
acuster: * Must attend weekly IRCs regularly and give advance notice when they cannot.
acuster: * Should maintain a released module.
acuster: * Must be willing to do whatever dirty work is necessary to keep the project moving forward.
acuster: o getting releases out on time (within a week of a request, or according to a schedule)
acuster: o making a representative available to the OSGeo foundation
aaime: I believe we qualify on all points
aaime: we mantain one module at least, work our ass of as anybody else (if not more) in terms of commits and proposals
acuster: except you don't maintain a released module and don't have time to do grunt work?
aaime: we do make releases
acuster: you see, we bend our schedule around geoserver
acuster: we bend our releases around it
acuster: which is great
aaime: ok, so, shall we remove the geoserver support and stop making releases, stop getting contributions?
acuster: but it does feel like geoserver does exactly and only the work it needs
acuster: jody does a bunch of work above and beyond what is needed for uDig
acuster: either we are all pitching in together
acuster: Martin dropping his work to answer you all's referencing questions
acuster: Jody, helping johann figure out styles in reference to what has gone before
acuster: i.e. a community
acuster: or we have no buisness working together
aaime: acuster, I've been answering Eclesia questions almost every day
acuster: ah, well good. That's the start of a community
aaime: I do review the proposal
Eclesia: (questions related to geoserver uses of styles interfaces)
aaime: I'm trying to help hbullen get a commit status (reviewin his patches)
aaime: I do answer questions on this channel on occasion
aaime: try to participate in the ml
aaime: if that's not good enough for you, pity, but if you want to kick me out of the PMC you'll have to ask for a PMC vote
aaime: last time I checked everyone of us had one vote
acuster: lol
afabiani: acuster what about code contribution? Is it not enough important for geotools as an OpenSource community?
desruisseaux: I would like to cool down the debate... This is true; Andrea has done a lot of work.
acuster: afabiani, no. Code is merely the beginning of a library.
desruisseaux: What I would like to stress out is just that its look like we are having different tagets
desruisseaux: (typo: targets)
aaime: desriusseaux, correct, it's true
cbriancon ha abbandonato la stanza (quit: "bye").
acuster: afabiani, try to figure out referencing. It's got the best javadocs in the whole project and no real docs.
acuster: afabiani, good luck.
desruisseaux: I would love to make GeoTools a library with the reputation of some C libraries
aaime: acuster, I tried twice, failed... probably I'm too stupid? (smile)
acuster: no, it's hard
afabiani: well anyone answers on matters he better know I guess
acuster: it's a whole field of science
acuster: except not entirely correct ie. a GeoTools datum is not really a datum
acuster: and needs a good command of affine transforms
acuster: and how those matrices are assembled
acuster: it's non trivial
acuster: the fact that no one apart from Martin understands it is not because everyone is stupid
acuster: it's hard and there's no good intro to the material
acuster: so code, even excellent code, is merely a start
acuster: if you want users and to build a community
acuster: that's why Jody is killing himself maintaining his wiki docs
acuster: okay, I'll shut up with one final line
aaime: acuster, sorry, but we're all different and we have different takes on how to do things
aaime: everyone tries to contribute in the way he sees fit
acuster: If GeoTools is going to be a library, we have a lot of serious work to do as a community and as a project.
aaime: you cannot pretend everybody else to contribute the way you want it
aaime: I want fast code, well unit tested, simple class structure, but I'm not going around asking people to help me in fixing gt2 slownesses
desruisseaux: The difficulty we are facing now as I see it, is that we are having different targets and we have a hard time to make those target compatibles.
aaime: I just ask if I can do it, and then I do it
desruisseaux: I would like to make a comparaison if you allow me...
aaime: and I don't consider gt2 necessarily a bad community because other people do not share the same core values as me
aaime: desriusseaux, please
desruisseaux: Sun got a lot of criticism because they came along with there java.util.logging package instead of adopting Log4J
desruisseaux: This is true that Log4J existed before, was popular, powerful, etc.
desruisseaux: But last time I looked at Log4J API, I saw a lot of classes and helper methods in a lot of packages.
desruisseaux: java.util.logging was supposed to be Log4J, but with only the "core" classes preserved (and renamed: Category --> Level), the remaining being left to third parties.
aaime: sure, but noting prevented Sun from becoming a contributor to the project and improve it... with the manpower they had, they could have done lots with it
desruisseaux: They got a lot of critic for that. Personnally, I loved this approach of extracting only the most essential part from a big library.
desruisseaux: The goal was not to improve Log4J; it was already good enough
desruisseaux: The goal was to extract the most essential part of it and let the remaining to third-parties.
aaime: the effect was to create confusion, piss off people, and in a way sentence the end of log4j
desruisseaux: It may have created some confusion for Log4J users, but it make the life easier for new users.
desruisseaux: I have the feeling that there is potentially much more new GeoTools users, if we can attract them with a library easier to browse.
aaime: Hum, did it? Today you start a project, most open source libraries are using commons logging, backed by log4j, and you wonder why there is an API in java that's not used... it seems it just added to the confusion? (smile)
desruisseaux: Well any way whatever java.util.logging was good or not is not the point I wanted to bring.
aaime: Ok, the point was?
desruisseaux: My idea is that Log4J was the bazar, with lot of classes and functionalities that Java logging don't have, while java.util.logging is a little bit more like a cathedral.
desruisseaux: Some like the bazar, thats fine for them
desruisseaux: Some other like the cathedral.
desruisseaux: I have the feeling that geoserver is more on the bazar side, while geomatys dream after the cathedral.
aaime: Sure thing, we're different
aaime: there is no doubt about it
Eclesia: (beside, log4j is hell for plugable systems using classloaders like netbeans platform, personal experience)
desruisseaux: The problem is that it is hard to build a cathedral when both the "cathedral" and the "bazar" groups are on the same SVN.
aaime: cathedral is usually more attainable with small development group, very focused, very strict rules and the like
desruisseaux: So my idea about GT3 was to exactly when I said in the "goal" section: not a new project, not a new API, but an attempt to extract a cathedral from the bazar.
aaime: desriusseaux, ok, but we may end up with a cathedral on one side and a bazar o the other... and after a while we'll forget we were once working togheter
desruisseaux: It depends if the API stay close enough
aaime: which depends on people willing to work on both sides
desruisseaux: As far as referencing and metadata are concerned, I do not envision any major structural change
desruisseaux: So I suspect that the merge would be relatively easy to performs between GT2 and GT3 for metadata and referencing.
desruisseaux: Things may be a little bit after geometry...
desruisseaux: (bit different)
acuster: bah, no
acuster: it will fit into the feature model fairly easily
desruisseaux: That would be good news.
desruisseaux: I think we are done for today?
aaime: yep
desruisseaux: Thanks for the discussion
desruisseaux: Bye all

  • No labels