xmlreader segfault on linux gcc-3.3.3 + fix

[G.o.D]

the xml reader seqfaults on linux with gcc-3.3.3 and maybe others

i am not sure why, maybe wchar_t isnt 16 bit anymore, but the
header of the textfile doesnt show up as 0xfffe or 0xfeff but as 0x3cfffe

you can fix it, changin line #80 (void CTextReader::convertToUTF16())
of CTextReader.cpp from
(header[0] == 0xFFFE || header[0] == 0xFEFF))
to
((header[0]&0xffff) == 0xFFFE || (header[0]&0xffff) == 0xFEFF))

btw:

thanx for irrlicht. this is the easiest api for a 3d engine ever seen, and
i have seen a lot, including commercial.

i am looking forward to see a big endian able version... i could offer my
help in porting if needed (also i have a veeery slow ppc machine, which
is not really fun to code c++ with )

[Tels]

the xml reader seqfaults on linux with gcc-3.3.3 and maybe others

i am not sure why, maybe wchar_t isnt 16 bit anymore,

AFAIK it is 32 bit under most Unixes.

but the header of the textfile doesnt show up as 0xfffe or 0xfeff but as 0x3cfffe

you can fix it, changin line #80 (void CTextReader::convertToUTF16())
of CTextReader.cpp from
(header[0] == 0xFFFE || header[0] == 0xFEFF))
to
((header[0]&0xffff) == 0xFFFE || (header[0]&0xffff) == 0xFEFF))

Thanx for the fix!

i am looking forward to see a big endian able version... i could offer my
help in porting if needed (also i have a veeery slow ppc machine, which
is not really fun to code c++ with :) )

There were already some people working on this, please search the forums. (I forgot where, when or who :)

Tels

[niko]

Ah, did not think that this could be the problem. Thanks for posting this

[G.o.D]

sorry, i didnt notice that this was only preventing the segfault, but the data of the xml file was completely ignored... so here is a fix, applyable with patch which finally makes the xml file reader reading the data (which i thought might be a good idea )

Code: Select all

--- orig.source/CTextReader.cpp 2004-03-04 16:25:36.000000000 +0100
+++ source/CTextReader.cpp      2004-03-14 09:52:53.000000000 +0100
@@ -77,14 +77,20 @@
        wchar_t* header = (wchar_t*)CData;
 
        if (CDataSize>=2 && 
-               (header[0] == 0xFFFE || header[0] == 0xFEFF))
+               ((header[0]&0xffff) == 0xFFFE || (header[0]&0xffff) == 0xFEFF))
        {
                // We've got UTF-16
 
-               TextData = &header[1];
+               TextData=new wchar_t[CDataSize];
+               for (u32 i=2; i<CDataSize; ++i) {
+                       TextData[i]=(wchar_t)(CData[i*2+1]<<8)|CData[i*2];
+               }
+               delete [] CData; 
+               CData = (c8*)TextData;
+
                TextDataSize = (CDataSize - 1) / (sizeof(wchar_t) / sizeof(c8)); // header 
 
-               if (header[0] == 0xFFFE)
+               if ((header[0]&0xffff) == 0xFFFE)
                {
                        // This is Big Endian, 
                        // convert to little endian.

[G.o.D]

yeah, i did it once again... the patch does not work correctly
so i now uploaded a final version which now works and is even tested this time...
here it is:

patch

you can call me microsoft