christian.muel...@nvoe.at ha scritto:
> @Andrea
>
> Hi Andrea, sorry to make some noise here. I misinterpreted the situation
> (bad coffee in the morning). I thought the shutdown method is primary
> intended for geoserver, obviously it is the other way around, making
> geotools ready for web ap
@Andrea
Hi Andrea, sorry to make some noise here. I misinterpreted the
situation (bad coffee in the morning). I thought the shutdown method
is primary intended for geoserver, obviously it is the other way
around, making geotools ready for web apps. This is a 100% +1 and I
hope we can end t
Simone Giannecchini ha scritto:
> Ciao a tutti,
> I guess that aaime wants clean up whatever is registered into generic
> registries, so jai operations and imageio reader, writers, instreams
> and outstreams. I am going to check aaime's patch afterwards.
>
> Anyway, andrea, (besides suggesting to
Ciao a tutti,
I guess that aaime wants clean up whatever is registered into generic
registries, so jai operations and imageio reader, writers, instreams
and outstreams. I am going to check aaime's patch afterwards.
Anyway, andrea, (besides suggesting to calm down :) ) I wanted to ask
you what we a
On 14 May 2010 18:25, Andrea Aime wrote:
>
> On GS devel there is a patch I provided that uses exactly that
> (attached, it's a bit messy but you should see the JAI cleanup
> spots)
>
Indeed. That's great - thanks very much Andrea !
Michael
--
Jody Garnett ha scritto:
> Cool - I am in favour of that.
>
> Is there any way we could "register" some activites for this hook? I
> expect that an OSGi context could also make use of them for a clean
> exit.
>
> For OSGi it is willing to startup / shutdown a module - we already
> have GeoTools i
Andrea Aime ha scritto:
> Well, the fact that it works just fine in desktop apps and does not
> work as fine in GeoServer would make me think otherwise.
> And you're well aware that there are other issues in J2EE in general.
> So much for being GeoServer only.
Taking a deep breath and trying a cal
Cool - I am in favour of that.
Is there any way we could "register" some activites for this hook? I expect
that an OSGi context could also make use of them for a clean exit.
For OSGi it is willing to startup / shutdown a module - we already have
GeoTools init so cleanup could be the spiritual o
On 14/05/2010, at 5:35 PM, christian.muel...@nvoe.at wrote:
> Reclamation :-)
>
> First, I use geotools also in an EJB Container,having its own life cicle and
> class loader. Second, I use geotools in a Java Applet, yet another life cyle.
> The Java VM is started only once and there can be many
Jody Garnett ha scritto:
> So a GeoTools.cleanup() then :-)
>
> Basically I would love to some how partition or set a checkpoint or
> something and cleanup; and keep on going ... Ideally I would love to
> do this on a Map by Map basis in uDig so when a Map is closed I can
> "flush" out and geotool
Michael Bedward ha scritto:
I'm following this with interest and am +0 about it so far because I
don't do web apps.
Regarding JAI, tthere is a method to unregister an operation
descriptor but I've never tried it. Simone: have you any experience
with this ?
On GS devel there is a patch I provi
christian.muel...@nvoe.at ha scritto:
> Reclamation :-)
>
> First, I use geotools also in an EJB Container,having its own life cicle
> and class loader.
>
> Second, I use geotools in a Java Applet, yet another life cyle. The Java
> VM is started only once and there can be many different applets
I'm following this with interest and am +0 about it so far because I
don't do web apps.
Regarding JAI, tthere is a method to unregister an operation
descriptor but I've never tried it. Simone: have you any experience
with this ?
Michael
--
So a GeoTools.cleanup() then :-)
Basically I would love to some how partition or set a checkpoint or something
and cleanup; and keep on going ...
Ideally I would love to do this on a Map by Map basis in uDig so when a Map is
closed I can "flush" out and geotools resources it was tracking.
We *d
Reclamation :-)
First, I use geotools also in an EJB Container,having its own life
cicle and class loader.
Second, I use geotools in a Java Applet, yet another life cyle. The
Java VM is started only once and there can be many different applets.
The only thing I want to say is that we should
christian.muel...@nvoe.at ha scritto:
> +1 for the concept.
> +1 for the name GeoTools.undeploy()
>
> -1 onWebApplicationUnload(), geotools itself has nothing to do with
> webapps,
> why should UDIG call onWebApplicationUnload() ??
It should not. That method is specifically designed for J2EE cle
+1 for the concept.
+1 for the name GeoTools.undeploy()
-1 onWebApplicationUnload(), geotools itself has nothing to do with webapps,
why should UDIG call onWebApplicationUnload() ??
IMHO it would make sense to have a GeoTools.redploy(). To give an example:
The imagemosaic-jdbc module could reread
Jody Garnett ha scritto:
> Thinking about that; I don't see a problem but kind of wish the
> shutdown was a bit more fine grained then that (ie datastore dispose
> cleaning up everything it is responsible for).
>
> I do like the idea of having the shutdown hook and us using it to
> catch leaks.
H
Thinking about that; I don't see a problem but kind of wish the shutdown was a
bit more fine grained then that (ie datastore dispose cleaning up everything it
is responsible for).
I do like the idea of having the shutdown hook and us using it to catch leaks.
Jody
On 14/05/2010, at 2:07 AM, An
Sounds like a great idea to me.
On 10-05-13 10:07 AM, Andrea Aime wrote:
> Hi,
> looking into why GeoServer leaks permgen memory on application
> redeploy in web containers I've found a number of problems in
> Geotools, mostly related to referencing and coverages: referencing
> keeps threads open
Hi,
looking into why GeoServer leaks permgen memory on application
redeploy in web containers I've found a number of problems in
Geotools, mostly related to referencing and coverages: referencing
keeps threads open and uses thread locals, the coverages subsystem
registers operations and similar obj
21 matches
Mail list logo