This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
Here's a scenario I ran into while working with adding of pathmap support to dbxgui. The Summary is a likley solution but I'm primarily interested in documenting the problematic scenario which would appear as a bug. Change development platform to a remote host of type, say, Solaris-Sparc using NFS. Build it, debug it ... alls cool. Now switch development to localhost and debug it ... things fail, pathmaps don't work, bpts won't work etc. Here's why. When building 'there' pathmaps will be automatically set up to map /export/home to /net/here/export/home. When the project is built there the working directory of the build will be /net/here/export/home/... This information is encoded in the compiler, fed to the debugger and thence to the IDE which pathmaps it to /export/home and accesses it here. Then, when you switch to localhost and debug you end up debugging the same executable that was built there dist/Debug/SunStudio-Solaris-Sparc/pctposixexport and has embedded in it paths of the from /net/here/export/home/... As a result breakpoints which are now on /export/home get rejected by dbx which expects paths to begin with /net/here/export/home. The problem will fix itself if one does a full rebuild 'here', but given that for debugging of large programs automatic "Build before run" is impractical there's a window where things will mysteriously not work. One possible solution is to have localhost have a redundant pathmap as well. It's the same pathmap which would be generated if files were being exported from 'here'. In a sense the machine 'here' becomes a remote machine and indistinguishable from any other remote machine. Another approach that may only work for dbx is to drop any leading / from paths sent to dbx. Then dbx will utilize a right-anchored best-match algorithm which was designed for helping differentiate between stop at main/file.c:50 stop at jill/project/file.c:20 But it's hard to evaluate all the implications of this.
*** Bug 180243 has been marked as a duplicate of this bug. ***
As I understand, if you open such project as /net/here/export/home/ (instead of just /export/home), everything works ok. Is this correct?
I don't think opening the project as /net/here will help. The problem is stale information encoded in the debugging information of a.out's and shared libraries which were built over there. Only a rebuild here will reset that information.