Community
Participate
Working Groups
If a plugin provides .options file, it is possible to enable tracing for this plugin via General > Tracing preference page. This bug covers adding the org.eclipse.jdt.core plugin to this preference page to make it easier to debug problems in runtime without restarting Eclipse.
I am already working on a patch for this and will push it to Gerrit when ready.
Gerrit review: https://git.eclipse.org/r/#/c/36504/
(In reply to Szymon Ptaszkiewicz from comment #2) > Gerrit review: https://git.eclipse.org/r/#/c/36504/ Patch looks good, I posted a new patch set with minor tweaks - extracted JavaCore.PLUGIN_ID + "/debug" into a constant. Szymon, let me know if this fine by you. Also, I am wondering if DEBUG_REGISTRATION.unregister() needs be guarded by a null-check. Or are we safe here?
(In reply to Jayaprakash Arthanareeswaran from comment #3) > (In reply to Szymon Ptaszkiewicz from comment #2) > > Gerrit review: https://git.eclipse.org/r/#/c/36504/ > > Patch looks good, I posted a new patch set with minor tweaks - extracted > JavaCore.PLUGIN_ID + "/debug" into a constant. > > Szymon, let me know if this fine by you. Looks good. Initially, I called it "DEBUG" instead of "debug" because in case we wanted to expose the value of the general JavaCore.PLUGIN_ID + "/debug" flag, we wouldn't need to rename the variable and we could just extract it to a field in JavaModelManager. But I guess that was kind of unnecessary forward-thinking in this case. Your change looks good to me. > Also, I am wondering if DEBUG_REGISTRATION.unregister() needs be guarded by > a null-check. Or are we safe here? It depends on whether we can get into state where o.e.jdt.core is half-started by finishing only part of the start(BundleContext context) method. From code analysis it looks that DEBUG_REGISTRATION will never be null because it's the very first thing in start(BundleContext context) so null-check is not strictly necessary. But we can add it to be on the safe side just in case if you want to.
(In reply to Szymon Ptaszkiewicz from comment #4) > (In reply to Jayaprakash Arthanareeswaran from comment #3) > It depends on whether we can get into state where o.e.jdt.core is > half-started by finishing only part of the start(BundleContext context) > method. From code analysis it looks that DEBUG_REGISTRATION will never be > null because it's the very first thing in start(BundleContext context) so > null-check is not strictly necessary. But we can add it to be on the safe > side just in case if you want to. That sounds reasonable to me. I have released it as is here: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=ec3141ffcff3411a0e4abbb098ed096a885f0d97
See bug 451902, comment #4 through bug 451902, comment #6.
Verified for 4.5 M4 using with build I20141209-2000.
