[Nexus] Maybe a bug?

[tieman]

I just upgraded some of my software to use NAPI 4.1.0.  I wanted to try 
and convert some of my HDF4 files to XML to see if XML is viable now.  I 
ran into a problem.

I read in an old HDF4 file that was generated with NAPI.  When I try and 
write the same file out with XML, the program crashes.  The program 
crashes at the line:

555>           memcpy(pPtr,data,dim[0]);

becuase dim[0] is some random huge number.  It appears that the previous 
call to NXXgetinfo which should have returned the proper dim size for 
the data fails at a subcall to analyzeDim when there is a character data 
field with only 1 character in it.  If the character field has more than 
one character, the typeString passed to analyzeDim looks like 
"NX_CHAR[x>1]" but if there is a single character, the typeString passed 
looks like "NX_CHAR" which causes analyzeDim to enter a switch statement 
based on the variable type--which is undefined in this case.

I could fix the problem by changing the order of lines 638-641 in 
nxxml.c from

638>      analyzeDim(attr,rank,dimension,iType);
639>      *rank = 1;
640>       *iType = NX_CHAR;

to

638>      *rank = 1;
639>       *iType = NX_CHAR;
640>      analyzeDim(attr,rank,dimension,iType);

so that type would be appropriately defined in analyzeDim. 

Not sure this is the right thing to do, but it got me past that problem 
and on to another one :)

Brian