Hi list,
I ran into a bit of a problem recently using the xmlsec library today on
a Linux x86_64 box. Here’s the situation:
1) The xmlsec library was compiled using the standard configure, make. (
--enable-debugging, --with-libxlst=no )
2) A piece of code that signs an xml doc was compiled and linked it to
the library.
3) The code involved a call to xmlSecDSigCtxSign()
result = ( xmlSecDSigCtxSign( dsigCtx, signNode ) >= 0);
This was failing because the struct in the library appeared to be
different then the one that our code was passing in. ( Memory was
shifted when it entered the xmlSecDSigCtxSign function ) Note: the code
works fine on 32-bit Windows and Linux.
We determined this was because of a compile flag that is set when
compiling the library: -DXMLSEC_NO_SIZE_T. (configure.in ln: 128)
This results in a size_t being an unsigned int rather than size_t .
(xmlsec.h line:43)
Apparently this is defined when configure detects a 64-bit system.
Obviously when we compiled our code we didn’t compile with this defined
causing a difference in the size of the struct relative to the piece of
code.
We can resolve this in two ways:
1) Not defining XMLSEC_NO_SIZE_T when compiling the library.
2) Compiling our application with XMLSEC_NO_SIZE_T defined.
Is anyone familiar with the implications of each of these actions?
Thank you kindly,
Nolan
--
Nolan Hurlburt
[EMAIL PROTECTED]
_______________________________________________
xmlsec mailing list
xmlsec@aleksey.com
http://www.aleksey.com/mailman/listinfo/xmlsec
