Re: Expat and UTF-16

Hi Erik,

How are you doing the compiles ? If you compile from the command line there is always an option to send the compiler listing to *PRINT, but that is not the default. Once the compile listing is in the iSeries spooled files, then normally you can see the line by line the errors shown under the line of code it complains about with letters indicating the exact point(s) in the line it is complaining about. So on the CRTCMOD command you need to do something like this CRTCMOD MODULE(xxx/xxx) SRCFILE(xxx/QILECSRC) OUTPUT(*PRINT) OPTIMIZE(10) DBGVIEW(*ALL) to see the errors in the listing. If you are using the IFS for source list the directory (this shows using normal iSeries source files in libraries). If you are using a program on your PC to send in the compiles to the server, normally there is some preferences area (depending on the tool you are using) to set up the CL command defaults that it will pass on to the server, and you'll have to update those.

But I think starting on V5R3 or V5R4 (can't remember which now), the compiler listing of the lines of code vs errors reported was off by random numbers of lines (ie it would show an error that didn't relate to the code written on that line at all and if you were lucky you could work out that it was a few lines above or below but often it was just really really off) ... complaints to IBM about this from me at that time did not result in fixes to that, so often to get around the problem I compiled back towards an older OS using the older compiler to find my error, then once fixed, recompiled on the current OS level !! I believe I encountered the problem at V5R4 and had to compile back to the lowest supported release to get a good listing but it could have been V5R3 that started the problem. I can't test now to be sure as I do not have access to a V5R4 or V5R3 box at the moment. To compile back to a certain release add the option TGTRLS(VxRxMx) to the
command where the 'x' are the release version mod that you want (it must be supported on your current release).

Take care,

Genyphyr Novak
iSeries Technical Consultant
France
+33 62 15 02 02





1. Expat and UTF-16 (Erik Olsson)


-----Inline Message Follows-----

I'm trying to use Expat with UTF-16 (ccsid 1200) encoded files.

Now, my issue is that I'm getting a compile error with 1 severe error, but
the compiler listing does not state exactly where that error occurs. The
Epilogue section at the bottom simply mentions that there is one severe
error, but not what nor where it is.
I currently have all relevant strings passed to Expat defined as wchar_t. If
I don't have XML_UNICODE and XML_UNICODE_WCHAR_T (which changes XML_CHAR
from char to wchar_t) defined I rightly get errors verifying that the
prototypes does not match, so that should be ok. But somewhere I have
apparently missed something which the compiler isn't keen on informing me
about.

Expat include and macros:
#include <libexpat/expat/xmlparse.c>
#define XML_UNICODE
#define XML_UNICODE_WCHAR_T

And the prototypes:

void start(void * userdata, const wchar_t * elem, const wchar_t ** attr);
void end(void * userdata, const wchar_t * elem);

Before I post the whole thing, has anyone done something similar with
success? And any ideas on how to get the error to be pointed out on the
compiler listing (never had this issue before)?

Thank you,
Erik