Re: [udig-devel] deadlock "fixed" .. release?

[andrea antonello <andrea.antonello@xxxxxxxxx>]

> The problem is entire mine! I did a terrible terrible job of that sentence…

:)


> Q: Is this bug reported in Jira yet?
>
> If so ... we can add a comment to note that this issue has existed for
> several previous versions.
>
> If not .. we can describe the issue (and how to reproduce it) and mention
> that it has existed for several previous versions.
>
> I hope that is more clear?

definitely! :)
Marco, can you please open a JIRA ticket? You did definitely the best
job in tracking down causes.


Jody, what did you mean with " Soundly like the know DnD issue is the
and a what is new doc is all we need."
I assume this refers to soemthing else?

Ciao,
Andrea




> --
> Jody Garnett
>
> On Sunday, 24 February 2013 at 6:09 PM, andrea antonello wrote:
>
> Hi Jody,
> it might be a language problem of mine, but I do not understand wnat
> the two statements below mean.
>
> Okay so I think we need is reported as not a new bug.
>
>
> We need to report it as a new bug?
>
> Soundly like the know DnD issue is the and a what is new doc is all we need.
>
>
> ??
>
>
> Ciao,
> Andrea
>
>
>
>
> --
> Jody Garnett
>
> On 22/02/2013, at 7:37 PM, Marco Foi <foimarco@xxxxxxxxx> wrote:
>
> I could track the "out of memory" issue in these versions
> 1.4.0 64bits
> 1.3.3 64bits
> 1.3.1 64bits.
> 1.2.2 64bits
>
> All behave exactly as the last version: they froze once javaw.exe process
> exceeds 512Mb.
>
> Marco
>
>
>
> 2013/2/22 Jody Garnett <jody.garnett@xxxxxxxxx>
>
>
> You can try out previous versions to check. It could simply be that we
> don't normally work with features this size?
> Still we should assemble what we know into a bug report.
> --
> Jody Garnett
>
> On Thursday, 21 February 2013 at 6:00 PM, andrea antonello wrote:
>
> @Marco, nice catch, thanks.
>
> I was able to arrive at the same out of memory issue (in the debugger,
> following your instructions).
>
> Solution? Hold on to previous geometry using a weak reference, and allow
> the "undo command" to disable itself if the geometry is no longer available?
>
>
> So you think that the freeze come from the OOM? That would be interesting.
> Is this something that changed recently or did we just never notice? I can't
> imagine that.
>
> Ciao,
> Andrea
>
>
>
>
>
>
> --
> Jody Garnett
>
> On Thursday, 21 February 2013 at 1:29 AM, Marco Foi wrote:
>
> Wow! Quite funny: I did some test the I started to write this mail
> -----
> I tested your workflow with 10m_admin_0_countries.shp from data_1_3.zip
> testdata and, still, cannot reproduce the deadlock.
> I tested it on the USA pacific coast area, near the boundary with Canada
> and I successfully edited the horizontal boundary between the two states
> (the line belonging to the USA polygon) many times, always cycling through
> unselect polygon (cycling in the Pacific).
> Just one silly doubt: to perform the test I am using the "Edit geometries"
> Tool. Is it right? You did not specify this but it seems to me the only one
> suitable for the task.
> Marco
> -----
> ..but before sending it, I reverted to uDig for a "last look" and...
> ..whoop! It frozed!
> In the picture you can(n't) see the Canada polygon on which I just
> clicked.. ..so something happened AFTER the original polygon was removed and
> BEFORE the higlighted one was drawn.
>
> So I made this little test: I just started switching selection from Canada
> to Alaska by clicking first on one polygon than on the ather and back
> again.. ..keeping an eye on javaw.exe memoy print.
> Here the result.
>
> 150mb : uDig Just started
> 168mb: new empty map created
> 205mb: 10m_admin_0_countries.shp added to map
> 260mb: click on Canada with "Edit Geometry" Tool
> 315mb: click on Alaska with "Edit Geometry" Tool
> 383mb: click back on Canada with "Edit Geometry" Tool
> 395mb: click back on Alaska with "Edit Geometry" Tool
> 432mb: click back on Canada with "Edit Geometry" Tool
> 441mb: click back on Alaska with "Edit Geometry" Tool
> 504mb: click back on Canada with "Edit Geometry" Tool
> 510mb: click back on Alaska with "Edit Geometry" Tool
> 514mb: click back on Canada with "Edit Geometry" Tool
> 514mb: click back on Alaska with "Edit Geometry" Tool -> uDig frozes!!
>
> So my idea is that "highlighted" polygons objects get in memory an there
> they get stacked.
> This stuff never gets garbage collected.. ..so we might have some serious
> memory leak problem here!
>
> Also this might be the real cause of the issue: not actually anything to
> do with editing itself.
>
> Marco
>
>
> 2013/2/20 andrea antonello <andrea.antonello@xxxxxxxxx>
>
> Hi Marco,
> could you please test with the uDig testdata set, which is how it happens
> to me?
>
> http://udig.refractions.net/files/data/data_1_3.zip
>
> I am using the countries admin 10m layer.
>
> Thanks,
> Andrea
>
>
>
>
> On Wed, Feb 20, 2013 at 1:44 PM, Marco Foi <foimarco@xxxxxxxxx> wrote:
>
> While using
>
> http://udig.refractions.net/files/downloads/branches/udig-1.4.0.win32.win32.x86_64.exe
> on a Win 7 64bit machine
> on the america_singlepart SHP, and doing the testing across the NORT
> AMERICA - SOUTH AMERICA boundary (to involve switching between to large
> polygons)
> I could NOT reproduce the lock!
>
> Marco
>
>
> 2013/2/20 andrea antonello <andrea.antonello@xxxxxxxxx>
>
> Hi Jody,
>
> I tried to reproduce the deadlock Andrea reported while using a debugger
> (in order to create my own better bug report). But was unable to catch it
> out ...
>
> Do you have any further tips for reproducing this Andrea?
>
>
> I am afraid not... let's summarize:
>
> - select USA polygon
> - move a node
> - unselect (click outside)
> - select another polygon
> - go back to the first
>
> Here I already am in freeze. Try to do it more and more.
>
> Is anyone else experiencing this deadlock?
>
> Ciao,
> Andrea
>
>
>
>
>
>
> --
> Jody Garnett
>
>
> _______________________________________________
> User-friendly Desktop Internet GIS (uDig)
> http://udig.refractions.net
> http://lists.refractions.net/mailman/listinfo/udig-devel
>
>
>
> _______________________________________________
> User-friendly Desktop Internet GIS (uDig)
> http://udig.refractions.net
> http://lists.refractions.net/mailman/listinfo/udig-devel
>
>
>
> _______________________________________________
> User-friendly Desktop Internet GIS (uDig)
> http://udig.refractions.net
> http://lists.refractions.net/mailman/listinfo/udig-devel
>
>
>
> _______________________________________________
> User-friendly Desktop Internet GIS (uDig)
> http://udig.refractions.net
> http://lists.refractions.net/mailman/listinfo/udig-devel
>
>
> _______________________________________________
> User-friendly Desktop Internet GIS (uDig)
> http://udig.refractions.net
> http://lists.refractions.net/mailman/listinfo/udig-devel
>
>
>
> _______________________________________________
> User-friendly Desktop Internet GIS (uDig)
> http://udig.refractions.net
> http://lists.refractions.net/mailman/listinfo/udig-devel
>
>
> _______________________________________________
> User-friendly Desktop Internet GIS (uDig)
> http://udig.refractions.net
> http://lists.refractions.net/mailman/listinfo/udig-devel
>
>
>
> _______________________________________________
> User-friendly Desktop Internet GIS (uDig)
> http://udig.refractions.net
> http://lists.refractions.net/mailman/listinfo/udig-devel
>
>
> _______________________________________________
> User-friendly Desktop Internet GIS (uDig)
> http://udig.refractions.net
> http://lists.refractions.net/mailman/listinfo/udig-devel
>
>
> _______________________________________________
> User-friendly Desktop Internet GIS (uDig)
> http://udig.refractions.net
> http://lists.refractions.net/mailman/listinfo/udig-devel
>
> _______________________________________________
> User-friendly Desktop Internet GIS (uDig)
> http://udig.refractions.net
> http://lists.refractions.net/mailman/listinfo/udig-devel
>
>
>
> _______________________________________________
> User-friendly Desktop Internet GIS (uDig)
> http://udig.refractions.net
> http://lists.refractions.net/mailman/listinfo/udig-devel
>