Re: [stellation-res] Question about Merging NPE

[Jeffrey Howard <jhoward@xxxxxxxxxxxxxxx>]

Hello,

I changed the two lines, but unforunately I'm still getting the
exception.  It's being thrown from inside the nearestCommonAnswer
method below the one which your fix of the == changes.  The signature
of the method that's actually throwing the NPE is 

	public BranchSpec nearestCommonAncestor(
		String projectName,
		BranchSpec leaf,
		BranchSpecSet targetParents)
		throws RepositoryException;

I think the exception is coming about because the call to
hc.getAllParents() shown below is returning null.  When nextParents is
used in the allParents.addAll call in the next line, that method
(BranchSpecSet.addAll) throws an NPE.

	BranchSpecSet nextParents = hc.getAllParents(parentsIt.next());
	allParents.addAll(nextParents);

Assuming I'm right in thinking that I didn't add any logging
statements that would change the line numbers, the
allParents.addAll(nextParents) call that's throwing an exception is on
line 1832 of the file.

Jeffrey





On Wed, Dec 25, 2002 at 11:26:10AM +0000, Mark C. Chu-Carroll wrote:
> Argh. I just took one look at the code, and it's an astonishingly stupid
> error. Downright idiotic to be truthful.
> 
> The code reads:
> 
> if (leafParents == null && leafParents.contains(target)) {
> 	return target;
> }
> 
> In other words, "if leafParents is null, please crash".
> 
> That's supposed to be a "!=".
> 
> I'm about 99% sure that just changing the two "=="s in that
> stretch of code to "!="s will fix it. I just did a reinstall of my
> system to a newer version of Linux, and I haven't finished rebuilding
> my environment, so I need to finish setting things up before I can test
> it. At this point, it'll be at least a day before I'm done with all
> the setup (I'm not able to work full time right now because I'm home
> with my daughter, and I need to fit in little bursts of works whenever
> she's occupied with something else. So things that would normally take
> only an hour or two of uninterrupted work are taking a lot longer.)
> 
> I'll make the change and test it as soon as I can. If you're in a hurry,
> just try changing the two statements that compare to null, and see what
> happens.
> 
> 	-Mark
> 	
> 
> 
> On Wed, 2002-12-25 at 15:59, Jeffrey Howard wrote:
> > On Wed, Dec 25, 2002 at 09:57:05AM +0000, Mark C. Chu-Carroll wrote:
> > 
> > > You've almost certainly found an edge case that my nearestCommonAncestor
> > > code doesn't cover. One thing that I'd like, just to make sure, is to
> > > see the sequence of commands your using to checkout, branch, checkin
> > > changes, and do the merge.
> > > 
> > > 	-Mark
> > 
> > Right.  The first time I ran across this, the branches, adds, and
> > checkins were all done by the Eclipse client.  I was only using the
> > command line client for the merge.  I started over with a new test
> > project using only the command line client.  All of these commands
> > were given with the options: "--location postgresql:stellation -u
> > jhoward -p password"
> > 
> > svc create testproject
> > svc co testproject
> > svc fork jhoward-sandbox
> > echo testfile > testfile
> > svc add testfile
> > svc checkin
> > svc -r main co testproject
> > svc merge jhoward-sandbox
> > 
> > At the final merge command I get the NPE thrown from within the NCA
> > method.
> > 
> > Jeffrey
> > 
> > _______________________________________________
> > stellation-res mailing list
> > stellation-res@xxxxxxxxxxxxxxx
> > http://dev.eclipse.org/mailman/listinfo/stellation-res
> -- 
> Mark Craig Chu-Carroll,  IBM T.J. Watson Research Center  
> *** The Stellation project: Advanced SCM for Collaboration
> ***		http://www.eclipse.org/stellation
> *** Work Email: mcc@xxxxxxxxxxxxxx  ------- Personal Email: markcc@xxxxxxxxxxx
> 
> 
> _______________________________________________
> stellation-res mailing list
> stellation-res@xxxxxxxxxxxxxxx
> http://dev.eclipse.org/mailman/listinfo/stellation-res