Oh, this bug isn't trivial to fix either - someone had a brain fart and
made pvfsEXELocation local to one file ... I'll let whoever's code that
is fix it.
Anyway, I think I found that bug. It was related to calling a
completion function that was expecting to decode a state machine control
block (the caller was passing the wrong pointer - apparently it was
declared with a void pointer argument so the compiler didn't catch it).
I tried greping around to look at that stuff, but since I don't know
all the places a completion function is called its hard to figure out if
they are all right. Anyway, we might just have to brute force it to
find any more. I'll get it all checked in for the next nightly.
Walt
Walter B. Ligon III wrote:
Rob,
there is a bug in test/client/vfs/test-mkdir at line 230. After making
and removing a couple directories the test decides to remake the
directory, but rather than call the create_directory function, creates
it itself and in the process forgets to check the pvfsEXELocation - thus
this fails if the test objects are not in the path (as on my machine). I
guess you have always had that in the path so it hasn't affected you.
Walt
--
Dr. Walter B. Ligon III
Associate Professor
ECE Department
Clemson University
_______________________________________________
Pvfs2-developers mailing list
[email protected]
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers
