Simon, you should seriously consider switching to a Linux machine for your data-intensive work. Any recent version of the Linux kernel has memory management, file systems and process schedulers that offer far better performance than Win32. I myself had one run-in with Windows. It was the end of 1994 and Windows 95 had just been released. I installed it, it crashed about a hundred times on me, I deleted it, switched to Linux and never had a reason to try my luck with Windows again.
Recently I had a look at Windows 7, because everyone talked about it so much. It took 10 minutes to boot up on my 5 year old PC which boots up Ubuntu in about 90 seconds. It then blinked and whistled for a long time before it came up with a terribly ugly GUI, popped up message bubbles all over the place and asked me to start some activation process. I deleted it. Nowadays, I use Windows to port software to it, because a lot of people are running Windows systems, so it needs to be done. Having said that, legacy applications are of course a problem if they represent heavy investment. But if the VM does not cut it, why not simply get a cheap laptop or find an old PC, put the Windows stuff on there and do the rest of the work on a Ubuntu PC? Ben ----- Original Message ----- From: "Simon Cropper (Botanicus Australia Pty Ltd)" <scrop...@botanicusaustralia.com.au> To: "Users and Developers mailing list" <gvsig_internacional@listserv.gva.es> Sent: Saturday, January 9, 2010 11:24:47 AM GMT +01:00 Amsterdam / Berlin / Bern / Rome / Stockholm / Vienna Subject: Re: [Gvsig_english] POTENTIAL ERROR gvSIG 1.9 (BN 1253) -- Jave Heap Error while editing table Ben, Having this error occur intermittently, I have not noticed that much of an increase in data being used or what I have asked it to do. That is, I don't see an increased workload as the culprit. That said this job is the largest GIS project that I have and requires the generation of multiple maps and low to moderately complex analysis. When using ArcView it to struggled. Personally I don't think it is the size of the project or data being manipulated I think it is the complexity of the tasks being requested. I get the impression, a please anyone out there who have tested gvSIG please don't take this the wrong way, most testing involves individual elements. Yes, you can add a layer. Yes, you can have labels based on two fields. Yes, you can categorize a field based on a particular field. Where I think I am pushing it is in asking gvSIG to juggle too many items - 33 ECW files (w/ scale criteria, put into group) and one moderately sized shapefile with ~300 geometries overing 100 kmĀ², multiple instances in the ToC with varying symbology and labeling strategies. It was only after the 4 instance of the same shapefile was inserted into the ToC and new labels were being inserted I ran into problems with memory. So the short response 'complexity' not 'size' appears to be the problem -- and following my analogy the juggler dropped a ball. Again it is debatable whether the program requested more memory than the VM could provide so a subroutine crashed when it could not create an array or whether the program has a fundamental flaw that has resulted in an error. If it was the latter I expect I would be able to reproducer the error - which I haven't. Therefore I think I am pushing the boundaries of gvSIG and the Java Machine it is contained within. I have knocked up my memory to 1000MB (half my memory) -- hopefully this will allow me to work through this project, which is nearly at an end. Interestingly I have just been offered an even larger mapping project covering a similar area. This will really be pushing the size limits. Interestingly it will be relatively simple. One shapefile, many ECWs, typology checking and attribute verification. I will post updates on how this goes. As a side note, you appear to have done work on Windows and Ubuntu. What are you currently using and which do you prefer? I am migrating towards Ubuntu but have stalled in replacing an in-house Visual Foxpro program that uses dual monitors (VM does not use dual monitors), so at present I am stuck as this program has 10-years of legacy programming that can not be readily replaced. Cheers Simon Simon Cropper Botanicus Australia Pty Ltd PO Box 160, Sunshine, Victoria 3020. P: 9311 5822. M: 041 830 3437. mailto: scrop...@botanicusaustralia.com.au <mailto:scrop...@botanicusaustralia.com.au> web: www.botanicusaustralia.com.au <http://www.botanicusaustralia.com.au> On 9/01/2010 8:53 PM, Benjamin Ducke wrote: > Hi Simon, > > I have tried setting Xmx> 1500 on a 4 GB WinXP machine and > got the same errors. On a Linux machine, I could easily increase > to> 2000 (stopped testing there). The reason is that the > JVM wants that much memory in one *contiguous* block and > Windows' memory manager is incapable of delivering that. > > Finding an optimal setting for all the Java VM memory settings > is something of a black art. Java tries to do all mem management > automatically, but that automatism is controlled by the interaction > of a number of options. > Even Sun seem to be struggling with the complexities of it: > > http://java.sun.com/performance/reference/whitepapers/tuning.html > > The error log you attached clearly shows that the VM ran out > of memory. Whether that is because it actually couldn't handle > the amount of data you threw at it or there is a bug in the > program code that leads to it trying to allocate too much > memory is a matter that needs looking into. > > Ben > > > ----- Original Message ----- > From: "Simon Cropper (Botanicus Australia Pty > Ltd)"<scrop...@botanicusaustralia.com.au> > To: "Users and Developers mailing list"<gvsig_internacional@listserv.gva.es> > Sent: Saturday, January 9, 2010 4:54:24 AM GMT +01:00 Amsterdam / Berlin / > Bern / Rome / Stockholm / Vienna > Subject: Re: [Gvsig_english] POTENTIAL ERROR gvSIG 1.9 (BN 1253) -- Jave Heap > Error while editing table > > Ben, > > 500M - gvSIG Opened OK > 800M - gvSIG Opened OK > 1000M - gvSIG Opened OK > 1300M - gvSIG Opened OK > - could open projects > 1500M - gvSIG Opened OK > 1501M - gvSIG Opened OK > 1600M - gvSIG Opened OK > 1700M - Error "Could not create Java machine" > 1800M - Error "Could not create Java machine" > 1600M - gvSIG Opened OK, plug-in services could not open > 1500M - gvSIG Opened OK, plug-in services could not open > 1300M - gvSIG Opened OK, could open project > > I would imaging this would be dependent on your available memory (2GB on > my system). I think I will set at 800M at present and see what happens > as I usually have more than my email client open when running gvSIG. > > Cheers Simon > > Simon Cropper > Botanicus Australia Pty Ltd > PO Box 160, Sunshine, Victoria 3020. > P: 9311 5822. M: 041 830 3437. > mailto: scrop...@botanicusaustralia.com.au > <mailto:scrop...@botanicusaustralia.com.au> > web: www.botanicusaustralia.com.au<http://www.botanicusaustralia.com.au> > > > On 8/01/2010 9:06 PM, Benjamin Ducke wrote: > >> Also note that under Linux you can set the heap space >> much larger than under Windows. In my experiments on Win XP, >> gvSIG would no longer start up if I set Xmx to> 1500 or so. >> Try how far you can go. >> >> Ben >> >> ----- Original Message ----- >> From: "Jorge Piera"<jorge.pi...@iver.es> >> To: "Users and Developers mailing list"<gvsig_internacional@listserv.gva.es> >> Sent: Friday, January 8, 2010 8:45:48 AM GMT +01:00 Amsterdam / Berlin / >> Bern / Rome / Stockholm / Vienna >> Subject: Re: [Gvsig_english] POTENTIAL ERROR gvSIG 1.9 (BN 1253) -- Jave >> Heap Error while editing table >> >> Hi. >> >> The optimum setting for memory depends on your environment and the >> application that your are running. I would try values from 500MB up to 1GB. >> >> Regards, >> Jorge. >> >> Simon Cropper (Botanicus Australia Pty Ltd) wrote: >> >> >>> Jorge, >>> >>> My settings are command = #JAVA# >>> -Djava.library.path="#GVSIG_INSTALL_PATH#\lib" -cp #CLASSPATH# -Xmx500M >>> -Xss1024k com.iver.andami.Launcher gvSIG gvSIG/extensiones #ARGS# >>> >>> What is the optimum setting for memory? I run a Windows XP SP3 2GB of >>> memory. >>> >>> Cheers Simon >>> >>> Simon Cropper >>> Botanicus Australia Pty Ltd >>> PO Box 160, Sunshine, Victoria 3020. >>> P: 9311 5822. M: 041 830 3437. >>> mailto: scrop...@botanicusaustralia.com.au >>> <mailto:scrop...@botanicusaustralia.com.au> >>> web: www.botanicusaustralia.com.au<http://www.botanicusaustralia.com.au> >>> >>> >>> On 8/01/2010 5:52 PM, Jorge Piera wrote: >>> >>> >>> >>>> Hello Simon. >>>> >>>> Hi. >>>> >>>> You can run gvSIG using more memory. Just edit the gvSIG.sh (or the >>>> gvSIG.ini) file located in the installation directory (folder bin) and >>>> find the text "-Xmx500M". 500M is the number of MB that you want to use >>>> when you execute the Java Virtual Machine. You can change this number by >>>> the number of MB that you want and next time that you run gvSIG, it will >>>> use this number of MB./ >>>> / >>>> Regards, >>>> Jorge. >>>> >>>> >>>> Simon Cropper (Botanicus Australia Pty Ltd) wrote: >>>> >>>> >>>> >>>> >>>>> Hi, >>>>> >>>>> I managed to generate a Java Heap error in gvSIG. >>>>> >>>>> Unfortunately I was unable to capture any error details. >>>>> >>>>> I can tell you what I was doing and hopefully you can reproduce. >>>>> >>>>> I was editing a shapefile's attribute table. Specifically I was changing >>>>> the values of a field used in complex labeling in individually defined >>>>> labeling classes - 10W = 10m high white writing, 5W = 5m high white >>>>> writing, etc. >>>>> >>>>> The shapefile is present 3 times in ToC. The first is site #, second >>>>> vegetation type, third quality. The label sizre is based on a field >>>>> called [LglGrp]. I was editing this field when I asked myself what the >>>>> previous instance of the shapefile looked like. Without really thinking >>>>> I clicked on the first instance (Site #) of the shapefile to view the >>>>> labels (without saving my changes) and BANG all hell broke loose - I had >>>>> numerous Java Heap Errors. >>>>> >>>>> I exited the program without saving and reentered. The shapefile was >>>>> intact but the recent edits were lost (obviously). I have scoured the >>>>> system but no error log was found. >>>>> >>>>> I have tried to repeat my steps but have not been able to triggered the >>>>> error again. >>>>> >>>>> Anyone know what would have triggered such a response? Memory issues? >>>>> >>>>> >>>>> >>>>> >>>>> >>>> >>>> >>>> >>> _______________________________________________ >>> Gvsig_internacional mailing list >>> Gvsig_internacional@listserv.gva.es >>> http://listserv.gva.es/cgi-bin/mailman/listinfo/gvsig_internacional >>> >>> >>> >>> >>> >> >> > _______________________________________________ > Gvsig_internacional mailing list > Gvsig_internacional@listserv.gva.es > http://listserv.gva.es/cgi-bin/mailman/listinfo/gvsig_internacional > > > > ------ > Files attached to this email may be in ISO 26300 format (OASIS Open Document > Format). If you have difficulty opening them, please visit > http://iso26300.info for more information. > > _______________________________________________ > Gvsig_internacional mailing list > Gvsig_internacional@listserv.gva.es > http://listserv.gva.es/cgi-bin/mailman/listinfo/gvsig_internacional > > > _______________________________________________ Gvsig_internacional mailing list Gvsig_internacional@listserv.gva.es http://listserv.gva.es/cgi-bin/mailman/listinfo/gvsig_internacional ------ Files attached to this email may be in ISO 26300 format (OASIS Open Document Format). If you have difficulty opening them, please visit http://iso26300.info for more information. _______________________________________________ Gvsig_internacional mailing list Gvsig_internacional@listserv.gva.es http://listserv.gva.es/cgi-bin/mailman/listinfo/gvsig_internacional