Sorry about the last patch, it was awful. Here is a new patch. I ran this version of the patch using check-xsddata-test-suite.py and ended up with no hexBinary failures and no leaks. I've also updated and included my hexBinary tests. BTW I think there are some problems with the base decimal type; it is only parsed into the lo value so hi and mi are always zero. I can fix it later but one question I have is what should we do on overflow condition (> 24 digits)? If a decimal string longer than 24 digits is processed, it may still be valid but we can't use the values in facet validation. I figure there are three choices: 1) print a warning and continue on. 2) return an error and possibly print a warning. 3) or set an overflow flag so that if the value is used later we can flag a warning. C. Bozeman On Thu, 2003-08-07 at 03:42, Daniel Veillard wrote:
On Tue, Aug 05, 2003 at 08:51:42PM -0500, Charles Bozeman wrote:This patch implements hexBinary values as strings. I've also included test code for validating hexBinary types.Hum, I tried to apply it and ran the regression tests again using check-xsddata-test-suite.py , it went from found 1035 test instances: 954 success 28 failures OK to found 1035 test instances: 954 success 28 failures Memory leak 33 bytes Apparently there is 3 memory leaks. Looking at the code it seems the string is allocated even if no data built is required, resulting in a memory leak for sure (if val == NULL , tmp is lost). There must be something else going on though... I will look at it. Daniel
Attachment:
xmlschemastypes.c-pat
Description: Text Data
Attachment:
hexbinary_0.xml
Description: Text Data
Attachment:
hexbinary_0.xsd
Description: Text document
Attachment:
hexbinary_1.xml
Description: Text Data