On 26 January 2011 09:15, Dave Page <dpage@pgadmin.org> wrote:
> We don't support non-unicode builds any more, so you might as well
> just inline the mb_str call where it's needed, and lose the rest.
Okay, I removed the entire function.
I had another look at our Flex/Bison problem. Look at this diff
output, after a fresh run of parser.sh:
diff --git a/pgadmin/include/pgscript/parser.tab.hh
b/pgadmin/include/pgscript/parser.tab.hh
--- a/pgadmin/include/pgscript/parser.tab.hh
+++ b/pgadmin/include/pgscript/parser.tab.hh
#include <string>
#include <iostream>
#include "stack.hh"
-#include "pgscript/pgScript.h"
-#include "pgscript/statements/pgsStatements.h"
-#include "pgscript/expressions/pgsExpressions.h"
-#include "pgscript/objects/pgsObjects.h"
-#include "pgscript/utilities/pgsContext.h"
#include "location.hh"
Restoring this by hacking the header doesn't quite fix the problem.
More headers must be hacked, where there doesn't appear to have been
headers removed. This may be symptomatic of a particular problem that
someone that really groks Flex++/Bison may be able to identify. What
do you think?
Regarding warnings, some cross-platform projects standardise on fixing
all warnings for a particular preferred platform. I thought you might
not appreciate fixes for warnings that you presumably don't even see
on your system (otherwise you'd have fixed them yourself), that crop
up quite a lot, on top of an already fairly extensive patch. I'll fix
them though.
BTW, pgscript developer docs indicate that Flex 2.5.33+ is supported,
so it obviously isn't that sensitive.
--
Regards,
Peter Geoghegan