Bugs item #3611503, was opened at 2013-04-21 10:34 Message generated for change (Comment added) made by michaudm You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=679906&aid=3611503&group_id=118054
Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: None Group: None >Status: Open >Resolution: None Priority: 5 Private: No Submitted By: Andrei_n (andrei-n1) Assigned to: Nobody/Anonymous (nobody) Summary: Create polygon from closed area tool freezes OJ Initial Comment: Open Jump stops working when using the Create polygon from closed area to generate a sequence of polygons. The OJ window will collapse and the program can only be closed from the Task Manager (I'm using Windows 7). ---------------------------------------------------------------------- >Comment By: michael michaud (michaudm) Date: 2013-04-26 11:40 Message: Hi Ede, Thanks for the help. Still not work with last change though I confirm this is not a .bat/.exe matter but a jre6 / jre 7 one ---------------------------------------------------------------------- Comment By: ede (edso) Date: 2013-04-26 03:13 Message: Andrei, please try rev 3532.. ede ---------------------------------------------------------------------- Comment By: ede (edso) Date: 2013-04-26 02:18 Message: > Oh, I get it, OJ freezes when launched from .exe, and not from .bat ! > Also, it does not exactly freezes, but do not response any more as if it > was waiting for a validation (looks like it is waiting for the OK cancel > button, but this one is not visible). tried on XP, jre8, run from bat - draw the hash symbol via four linestrings - use the tool - get the not found/search OK/cancel dialog - click OK thereafter ithe java process does not hang! it redraws perfectly when brought back to front. you can also kill it via Strg+C in the console. but it is modally deactivated. there must be a modal dialog linger around invisibly or off the screen blocking access. i can confirm, jre6 does not sport these symptoms. ..ede ---------------------------------------------------------------------- Comment By: Jukka Rahkonen (jratike80) Date: 2013-04-26 01:46 Message: Situaltion is muchj worse with jre 1.7 than with 1.6. I rechecked the memory consumption loop behaviour and and at least the GC loop is much faster after using the Create polygon tool than before using it. I forgot to say that I do not do anything with OJ but it just stays idle from my point of view. Another thing: with jre 1.6.0_33 OpenJUMP keeps on reaction for me but with jre 1.7.0_21 OJ stops reacting totally after using the tool on a non-closing area. Test dataset is easy to create by digitizing four lines in a form of # If I use the tool inside the unnnoded hash and click OK the jam happens immediately. If I press Cancel the jam does not happen untid I try to do the same for the second time. ---------------------------------------------------------------------- Comment By: michael michaud (michaudm) Date: 2013-04-26 00:08 Message: Jukka, I think the memory usage you observed is the normal behaviour of the garbage collector. As long as there is no absolute need to free memory, GC collects garbage in a way which is quite difficult to predict (at least for me). About the bug, I must admit that I wrapped the process in a thread (which is unusual for an editing tool) and that the way the tool, the progress bar and the ok cancel panel interact is quite complex and may still have some bug. It is very strange, however, that it works fine from the bat and that it blocks from the exe. ---------------------------------------------------------------------- Comment By: Jukka Rahkonen (jratike80) Date: 2013-04-25 23:28 Message: I made a test too. I digitized some unnoded lines to make a ladder like figure like in the video. Then I used several times the Create polygon from closed area and canceled operation each time. I do not have crash and no messges appear into the Java console. However, I can see from the OJ panel that the reserved memory is changing continuosly. The map is almost empty so the memory usage is low, but the numbers change 14->15->16->17->18->19->20->21->22 MB and then back to 14 MB and a new round starts. This loop continues forever. OpenJUMP 1.6.1 relaase, Windows Vista 32-bit and jre 1.6.0_27 32-cbit Oracle. OJ started through the oj_windows.bat. ---------------------------------------------------------------------- Comment By: michael michaud (michaudm) Date: 2013-04-25 23:10 Message: Yes you're right. I thought the jvm crashed after the second click on cancel, but it just hides the application windows. and... Oh, I get it, OJ freezes when launched from .exe, and not from .bat ! Also, it does not exactly freezes, but do not response any more as if it was waiting for a validation (looks like it is waiting for the OK cancel button, but this one is not visible). ---------------------------------------------------------------------- Comment By: ede (edso) Date: 2013-04-25 19:25 Message: mmmmh, Michael how do you figure that it's a jvm crash? the OJ gui is still there but unresponsive it seems. Andrei: please edit your oj_windows.bat and remove the 'rem' in front of 'set JAVA_BIN=java' rem -- uncomment to use 'java' for console output, if unset defaults to 'javaw' for background jre -- rem set JAVA_BIN=java redo the steps of your video and attach the full content of the terminal window that stays open here.# _also_ please try the lastest jre7, which is update 21, just to be sure it is no java bug that has already been fixed. _also_ how much physical RAM do you've got? ..ede ---------------------------------------------------------------------- Comment By: Andrei_n (andrei-n1) Date: 2013-04-25 16:00 Message: So in the end it's a fancy jvm crash and not an OJ bug. Should I address this issue on a java bug tracker then? Here's the info from the OJ about panel: Java version - Java HotSpot(TM) 64-bit Server VM 1.7.0_07 (amd64) OS - WIndows 7 (6.1) Max memory - 7.9 GiB Total memory 366.2 MiB Comitted memory 75.6 MiB I can't find the hs_err_pidxxxx.log file. Where should I search for it? ---------------------------------------------------------------------- Comment By: michael michaud (michaudm) Date: 2013-04-25 11:47 Message: Wow, seems different from what I get. Yours is a jvm crash. Impressive. Can you give more information about environment (infos from OJ info panel for ex.). Also, a jvm crash usually produces a log file called hs_err_pidxxxx.log where xxxx is the pid number of the process Not sure it helps, but better than nothing. ---------------------------------------------------------------------- Comment By: Andrei_n (andrei-n1) Date: 2013-04-25 04:51 Message: Hi Michaël, Unfortunately my bug wasn't fixed in r. 3530. I created a short screen capture film with the OJ crash: https://www.wetransfer.com/downloads/68d160c947902e20729799b9a0f75ee020130425114811/de362af00d3c090c84e2d0e22732135020130425114811/8fbf55 ---------------------------------------------------------------------- Comment By: michael michaud (michaudm) Date: 2013-04-24 13:34 Message: Hi Andrei, I tried a fix, but could not reproduce the bug, even before the fix. Let me know if according to your test, r3530 fix the problem. ---------------------------------------------------------------------- Comment By: Andrei_n (andrei-n1) Date: 2013-04-23 14:38 Message: I'm now pretty sure there was no change which caused this error in rev. 1.5.1 or 1.5.2. I think I didn't get this error back then simply because I rarely used the tool. Your error behaves exactly like mine. I think it is related to my bug. In my case, after the first OJ collapse the error will reappear almost every time (just tested version 1.6.2 and the problem is still there). ---------------------------------------------------------------------- Comment By: michael michaud (michaudm) Date: 2013-04-23 13:02 Message: still can not reproduce the problem with your data and your instructions. Can you reproduce the problem with the current version (Nightly Build) ? The last fix on this plugin has been made a few days before 1.5.2 release, which means there is no reason that the plugin behaviour changed after 1.5.2. Oh, after many tests, I could get the following error. Now I get this error almost each time I click. But the window does not collapse. This is probably a concurrent issue. Don't know if it can be related to your bug. java.lang.NullPointerException at org.openjump.core.ui.plugin.edittoolbox.cursortools.FillPolygonTool.gestureFinished(FillPolygonTool.java:124) at com.vividsolutions.jump.workbench.ui.cursortool.AbstractCursorTool.fireGestureFinished(AbstractCursorTool.java:482) at com.vividsolutions.jump.workbench.ui.cursortool.MultiClickTool.finishGesture(MultiClickTool.java:248) at com.vividsolutions.jump.workbench.ui.cursortool.MultiClickTool.mouseReleased(MultiClickTool.java:139) at com.vividsolutions.jump.workbench.ui.cursortool.LeftClickFilter.mouseReleased(LeftClickFilter.java:93) at com.vividsolutions.jump.workbench.ui.cursortool.DelegatingTool.mouseReleased(DelegatingTool.java:105) at java.awt.AWTEventMulticaster.mouseReleased(AWTEventMulticaster.java:273) Michaël ---------------------------------------------------------------------- Comment By: Andrei_n (andrei-n1) Date: 2013-04-23 01:12 Message: 1. The problem is not related to the geometries. It is related to the warnings which appear when an area intended for polygon fill is not closed. 2. It happens after hitting the warning "cancel" button 2 or more times. After closing and restarting OJ, it will appear even when all the areas you want to fill are closed areas (the problem not dependent anymore on the appearance of the warnings) 3. I don't think it's necessary to send a dataset. Any simple dataset will do, even a layer with only 3 objects. You just have to try to fill a non-closed area several times and press cancel to the warnings. Just in case, here's a simple dataset: https://www.wetransfer.com/downloads/174d44f3cafe9fc6ac0772b02ad55d5220130423080815/2c9b65f165fbad1af50813ca9684d31520130423080815/5356f3. You can try filling any of the 3 bounded (but not closed) areas. 4. No, there;s no relation with the speed of the button click. I hope this helps. Please tell me if you need further information. ---------------------------------------------------------------------- Comment By: michael michaud (michaudm) Date: 2013-04-22 23:46 Message: I still could not reproduce it. Here are a few more questions : Do you think the problem is related to the geometries ? Does it happen with one polygon to fill only or ni several cases ? I would be interested in the dataset which makes the application freeze. Have you many features displayed when you click in the view (long task with the progress bar appearing on screen) ? Can you reproduce it with a small dataset ? Do you think that the bug is in relation with when you click on the button (clicking fast, before the previous task ends) or can you reproduce it when you click slowly ? Thanks for your help ---------------------------------------------------------------------- Comment By: Andrei_n (andrei-n1) Date: 2013-04-22 15:40 Message: I forgot that only noded intersections are considered for determining the closed area. So the area from the screenshot below is actually correctly hinted as not being closed. But the relation betwen the warnings and the eventual OJ crash remains valid. ---------------------------------------------------------------------- Comment By: Andrei_n (andrei-n1) Date: 2013-04-22 15:32 Message: I was happy too soon :( The problem is preceded by a ''clicked area is not closed'' warning. This error appears even when I want to insert a triangle-type polygon in a closed area. See this screenshot: http://img833.imageshack.us/img833/9694/polygonerror.png After retrying to insert the triangle several times and hitting cancel to the warnings OJ will freeze. Of course this also happens after genuine warnings are cancelled. Afterwards OJ will always freeze when generating several consecutive polygons from closed areas. The only solution is to reinstall the program. ---------------------------------------------------------------------- Comment By: Andrei_n (andrei-n1) Date: 2013-04-22 14:42 Message: This is funny. I don't encounter this error anymore in the latest snapshot (rev. 3525). I think you fixed it Michael. Perhaps it was related to some other bug (?) It happened always after drawing more than 3-6 polygons. The 1.6.0RC1 release featured this issue, but some older version (either 1.5.1 or 1.5.2) was problem-free. ---------------------------------------------------------------------- Comment By: michael michaud (michaudm) Date: 2013-04-22 14:06 Message: Sorry, I've tested with several datasets and jre... but I cannot reproduce the problem. Is this a problem you encountered just once or did you get it several times ? ---------------------------------------------------------------------- Comment By: Andrei_n (andrei-n1) Date: 2013-04-21 11:25 Message: Sequence of polygons means creating polygons from closed area in a quick succession. There's no error message and I can't obtain any log output. The OJ window is suddenly minimized and the program is frozen. ---------------------------------------------------------------------- Comment By: michael michaud (michaudm) Date: 2013-04-21 10:49 Message: What do you mean by a sequence of Polygons ? Have you a stacktrace or a logfile ? ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=679906&aid=3611503&group_id=118054 ------------------------------------------------------------------------------ Try New Relic Now & We'll Send You this Cool Shirt New Relic is the only SaaS-based application performance monitoring service that delivers powerful full stack analytics. Optimize and monitor your browser, app, & servers with just a few lines of code. Try New Relic and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_apr _______________________________________________ Jump-pilot-devel mailing list Jump-pilot-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel
