Community
Participate
Working Groups
Created attachment 94500 [details] proposed patch This bug-report is an excerpt from https://bugs.eclipse.org/bugs/show_bug.cgi?id=217292 The proposed patch fixed the problems with the fast machine, but not the original bug reported, therefore i created this entry. Description: The problematic steps were that in Java_org_eclipse_cdt_utils_spawner_Spawner_exec0 a thread for watching the process termination was started. It seemed that this thread couldn't get the correct handle for the process to wait for. Therefore it terminated and for the spawner treated this as if the process couldn't be started. The patch I attached uses the no thread for the startup check, e.g. it waits either for the waitEvent, which is triggered from starter.exe and indicates that everything went ok, or for the process termination which indicates that the starter.exe wasn't run correctly. 2nd part of the thread was to do the proper cleanup, which is still handled by the thread, but it's only launched if everything started up correcly. When passing the args to the thread procedure I noticed that some memory might get screwed up, so I explicitely handled this.
Is the dll attached to bug 217292 still valid?
Yes it is still valid. But I don't know the original CDT build environment (if there is a dedicated one) that is used to build this for official releases.
Created attachment 94698 [details] spawner.dll built with MinGW The patch looks good to me. I have built the dll with the Makefile from CVS using - gcc version 3.4.2 (mingw-special) - JDK 1.4.2_14
Committed to HEAD and cdt_4_0. Thanks for the patch!
*** Bug 217292 has been marked as a duplicate of this bug. ***
