Hi there,
I just had to revert:
Replace suitable equalsAscii calls with equalsAsciiL.
diff --git a/oox/source/ole/axbinaryreader.cxx
b/oox/source/ole/axbinaryreader.cxx
index 36cc2d7..105db2b 100644
--- a/oox/source/ole/axbinaryreader.cxx
+++ b/oox/source/ole/axbinaryreader.cxx
@@ -150,7 +150,7 @@ bool AxFontData::importGuidAndFont( BinaryInputStream&
rInStrm )
OUString aGuid = OleHelper::importGuid( rInStrm );
if( aGuid.equalsAsciiL( RTL_CONSTASCII_STRINGPARAM(
"{AFC20920-DA4E-11CE-B943-00AA006887B4}" ) ) )
return importBinaryModel( rInStrm );
- if( aGuid.equalsAscii( OLE_GUID_STDFONT ) )
+ if( aGuid.equalsAsciiL( RTL_CONSTASCII_STRINGPARAM( OLE_GUID_STDFONT ) ) )
return importStdFont( rInStrm );
return false;
}
This sort of thing.
The code is not correct; OLE_GUID_STDFONT is a const sal_Char * pointer
- so when we start doing sizeof() on it to detect the string length it
gets a -very- wrong (and dangerous) value. Luckily Caolans' sal patch
breaks the build if you have a recent enough compiler for this. Of
course; quite -why- it is a const sal_Char * I have no idea - it betrays
a lack of faith in the compiler's ability to optimise duplicate string
instances I suspect - prolly worth auditing their use, and fixing that
too. There was some madness before about sharing the same string
instance across modules via. a symbol which just burns space vs. having
it duplicated in two libraries; we should prolly undo that and just use
some defines.
I thought we had ~nearly finished this hack - of fixing up all the
equalsAscii's that can be safely done :-)
Thanks,
Michael.
--
michael.me...@novell.com <><, Pseudo Engineer, itinerant idiot
_______________________________________________
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice
