Community
Participate
Working Groups
There are 3 issues: 1. The install size that is gathered by the update wizard's TargetPage does not recursively get the size of child features. Currently only the parent feature and its child features are accounted for in the size. I have created a patch for this issue. The fix accounts for children's children. Changes made to SiteFile.getInstallSizeFor(IFeature). Now if there is a feature who's heirarchy is the following, the correct size info will be collected: Feature1 includes Feature2 Feature2 includes Faeture3 Feature3 includes Feature4 etc. 2. The download size information for feature's plugins and non-plugin data is not used anywhere. I have created a patch for this issue. The patch first fixes how the download size is retrieved in SiteFile.getDownloadSizeFor(IFeature) (changes based on changes made to SiteFile.getInstallSizeFor(IFeature)). The patch then uses the download size information in the update wizard's TargetPage. The page now adds together the download size and the install size in order to create the "required" size. Since all plugins and non-plugin data are downloaded before being installed, the download size should be incorporated into the "required" size: required = downloadSize + installSize. 3. The update wizard's TargetPage allows the user to continue even if the "require" size is greater than the "available" size. I have created a patch for this issue. The patch tests the see if the "required" size if greater than the "available" size. If it is greater, the page is set to be "not complete". There is a TODO in the code that says we need to add/set an error message telling the user what the issue is (not enough disk space). I will attach the patch info soon.
Created attachment 36364 [details] update core patch for download size and install size
Created attachment 36365 [details] update ui patch for download-size and install-size
committed
Please re-open and check bug 136907
Doug, you must account for platforms like Mac where we don't have native code for computing avaiable disk space (-1 returned).
Just for fun, when pack200 versions of JARs are availalbe they will be way smaller than the JAR sizes reported in the feature files.
Created attachment 38848 [details] Fix for systems where we cannot get available space I have coded a patch, but I do not have an environment to test it in. The patch is very simple: //if the available space was retireved from the OS and the required space is greater that the available space, do not let the user continue if(available != LocalSystemInfo.SIZE_UNKNOWN && required > available){
This looks good at first glance.
I tested the patch on 32RC1 on a powerbook g4 12" running MacOSX 10.4.6. The patch solves the problem :-) Thanks
*** Bug 136907 has been marked as a duplicate of this bug. ***
Thanks - released.
*** Bug 137715 has been marked as a duplicate of this bug. ***
*** Bug 73097 has been marked as a duplicate of this bug. ***
*** Bug 138116 has been marked as a duplicate of this bug. ***
*** Bug 138225 has been marked as a duplicate of this bug. ***
*** Bug 138228 has been marked as a duplicate of this bug. ***
*** Bug 138266 has been marked as a duplicate of this bug. ***
*** Bug 138548 has been marked as a duplicate of this bug. ***
*** Bug 138592 has been marked as a duplicate of this bug. ***
I don't know if it is really related to this bug or rather the fix, but now, using the latest nightly (need that fix!) from 2006-04-26, Update crashes if I invoke it from the menu: 2006-04-27 09:29:38.166 eclipse[3264] *** -[NSCFType length]: selector not recognized [self = 0x307150] 2006-04-27 09:29:38.166 eclipse[3264] *** Uncaught exception: <NSInvalidArgumentException> *** -[NSCFType length]: selector not recognized [self = 0x307150] Apr 27 09:29:46 grexmac crashdump[3267]: java crashed Will attach a crash log.
Created attachment 39636 [details] Eclipse crash on Update/Install
Steve, another 'interesting' crash that seems like it is coming from SWT land.
The stack looks a bit funny. Has the application got to it's event loop? The bottom of the stack doesn't contain the Mac calls that readAndDispatch() makes. Is the crash coming from application code that is calling Control.setCursor() before a event loop is running?
Does the crash happen every time?
yes, every time, right after invoking the menu and before the dialog even comes up. I see the "beachball of death" spinning for a while before I get the JVM-crash. Error number differed however. Trying now with RC2 (not 2a)...
I was not able to reproduce this crash on Mac/Intel 10.4.6 or Mac/ppc 10.4.6 running this build (I20060427-0800) and this VM ssq% java -version java version "1.5.0_06" Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0_06-110) Java HotSpot(TM) Client VM (build 1.5.0_06-61, mixed mode, sharing)
Using 3.2rc1 (I wrote rc2 above, which is not yet out..), Eclipse crashes right on startup: Apr 27 19:08:47 grexmac /usr/bin/java: An unexpected Java error has been detected by HotSpot Virtual Machine. Apr 27 19:08:47 grexmac /usr/bin/java: If this error is reproducible, please report it with the following information: Apr 27 19:08:47 grexmac /usr/bin/java: 1. Provide the steps to reproduce, a test case, and any relevant information Apr 27 19:08:47 grexmac /usr/bin/java: 2. The corresponding JavaNativeCrash_pid<num>.crash.log (Java state) Apr 27 19:08:47 grexmac /usr/bin/java: 3. The corresponding <name>.crash.log (native state; generated by CrashReporter) Apr 27 19:08:47 grexmac /usr/bin/java: 4. This data: Apr 27 19:08:47 grexmac /usr/bin/java: Java VM: Java HotSpot(TM) Client VM (1.5.0_06-57 mixed mode) Apr 27 19:08:47 grexmac /usr/bin/java: Bus Error (0xa) at pc=0xfffeff20 Apr 27 19:08:47 grexmac /usr/bin/java: Process ID: 3418, Current Thread: 25184768 Apr 27 19:08:47 grexmac /usr/bin/java: File report at: http://bugreport.apple.com/ Apr 27 19:08:47 grexmac /usr/bin/java: An error report file has been written to: Apr 27 19:08:47 grexmac /usr/bin/java: /Users/grexe/Library/Logs/Java/JavaNativeCrash_pid3418.crash.log Apr 27 19:08:51 grexmac crashdump[3423]: java crashed Apr 27 19:08:58 grexmac crashdump[3423]: crash report written to: /Users/grexe/Library/Logs/CrashReporter/java.crash.log This indicates that my Java-installation is corrupt for some reason, or am I mistaken? Maybe I should apply the 4.6 combo update again? (I have problems with Java applets in browsers, too) If this is true, I'm sorry for the confusion. Will attach the Eclipse-log and keep you updated.
Created attachment 39706 [details] Java crash log of Eclipse crashing on start
It looks like the Mac free space issue means that I can't install a number of plugins under 3.2RC1: oXygen and Subclipse for example. Is there a workround for this, or do people with Macs have to revert to 3.2M6 until 3.2RC2 comes out?
RC2 is supposed to be built sometime today - can't you hold on for just a bit longer?
Thanks Dejan, I hadn't realised we were that close to RC2. You're right, that does make the problem go away...
hi all. I also get this issue. does it fix in Version: 3.2.0(Build id: M20060629-1905)?
This is still occuring using Eclipse 3.2 final and Microsoft Vista Beta 2 (32bit). Is there a workaround?
bug is still in Eclipse 3.2, Win 2003
we do not officially support neither 2003 or xp-64 or vista. However, we are aware of the problems and there are bugs (bug 142815) open for this OSes. Fix was committed for 2003 which should also fix xp-64. We are hoping that it does fix vista too, however we can not be sure. So people that want to help, can download our newest nightly build and try it (it should be also in yesterdays integration build). Thank you all.