Link failure in src/import-export/test/test-link.o

Daniel Vetter daniel at ffwll.ch
Mon May 15 17:23:33 EDT 2006


On Mon, May 15, 2006 at 03:17:17PM -0400, Josh Sled wrote:
> On Mon, 2006-05-15 at 19:04 +0200, Daniel Vetter wrote:
> > On Sun, May 14, 2006 at 07:16:29PM -0400, Derek Atkins wrote:
> > > Hmm, I can't reproduce this with r14062.
> > > 
> > > -derek
> > 
> > Tracked it down with binary search on the commits. Breakage was introduced by
> > the following change:
> 
> I can't (and haven't been able to) reproduce this either.
> 
> I'd believe it's more a clean-compilation issue than actual breakage.
> Can you reproduce after `make clean`ing and re-building from SVN HEAD?
> If not, what about `make distclean`?

I did a new svn co && ./autgen.sh && ./configure && make on svn head (rev
14077), which failed, too. This should exclude some random hardware error on my
side (which didnt yet happen with this machine here, anyway - just to check).
For the binary search I did a svn-clean && ./autogen.sh && ./configure && make
each time. 

As already told, this revealed that rev 14037 introduced the breakage for me.
This change added the function the linker stage complained about, so it all
makes kinda some sense - beside that I have no clue what the real cause is.

All revision I tested from 14037 on have the same error, therefore it doesnt
look like something random either. I rechecked rev 14033 too, again doing
svn-clean beforehand, to cancel out some random debian unstable upgrade glitch,
but it compiles and runs as flawlessly as a few days ago.

Right now I've done a svn up (now at rev 14078) then

svn diff -r 14036:14037 | patch -p1 -R

this compiles cleanly and runs flawlessly.

Just some random datapoints about my system, if this helps perhaps:
gcc version 4.0.4 20060507 (prerelease) (Debian 4.0.3-3)
GNU ld version 2.16.91 20060413 Debian GNU/Linux
all on a amd64 box

Is there anything else I can check/provide?

-daniel


More information about the gnucash-devel mailing list