Re: NEC-LIST: Could this be a NEC2 bug?

Hello Neoklis,

Difficult to say where things go wrong, without a deeper look
and/or further debugging.
With a very short look at it, it seems to me that global array
ICON1(j) already contains an 'illegal' value, delivered from
outside the CMWW routine. In Fortran one can choose to enable
or disable array bounds checking. Maybe array checking by
default was turned off, thus resulting in no particular error
when running the supplied Nec file. When porting to C(++) the
'error' shows up.
I can't remember anymore what settings I used when compiling/
linking my Nec2dXS executable files.

Arie.

> While developing a multi-threading version of "xnec2c", the
> GUI version of "nec2c" (my translation to C of NEC2), I had
> segfault crashes of the app when using the EK card (extended
> thin wire kernel) in a structure that has a wire element
> connected to a patch. The same problem exists in nec2c which
> is nearer to the original FORTRN NEC2 so I set off to trace
> it and found that an index to the ICON2 segment connection
> array was out of range by 10000, which it the number added
> to entries in this array to indicate a connection to a patch.
>
> In order to make a comparison with NEC2 I added printf/WRITE
> statements to print out the value of the index to ICON2 in
> both the original FORTRAN version and my translation. I found
> that both programs printed out the same values for ICON2
> array members, one of which was offset by +10000. But
> according to the FORTRAN NEC2 PARAMETERS statement the
> dimensions for ICON1 and ICON2 are 2*MAXSEG where MAXSEG=1500
> in the original nec2dx.f code.
> etc.

-- 
(Please ignore the advertisements below)Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen! 
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
-- 
The NEC-List mailing list
NEC-List_at_robomod.net
http://www.robomod.net/mailman/listinfo/nec-list