Hi,
Crashing on Windows 7 has been reported [1] and confirmed. From IRC
discussions it appears that this might be a CRT issue. It seems this bug
does not happen on (windows < 7 and 32 bit).
[1] http://bugzilla.abisource.com/show_bug.cgi?id=13109
I have attached a test program and makefile for a very simple redland
client which should produce the output given below.
win7-crt-redland]$ ./main
serializer:0x11b5750 data:0x11b98e0 data_sz:269
data follows...
<?xml version="1.0" encoding="utf-8"?>
<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
<rdf:Description rdf:about="http://www.pants.com/">
<ns0:position
xmlns:ns0="http://www.pants.com/current/">On</ns0:position>
</rdf:Description>
</rdf:RDF>
We did not crash! exiting normally now...
win7-crt-redland]$
If somebody with windows 7 access and the inclination to test this
little main.cpp could that would be wonderful.
If this is a CRT issue, there are these from rdf_internal.h in redland:
#if defined(LIBRDF_MEMORY_SIGN)
#define LIBRDF_SIGN_KEY 0x04Ed1A7D
void* librdf_sign_malloc(size_t size);
void* librdf_sign_calloc(size_t nmemb, size_t size);
void* librdf_sign_realloc(void *ptr, size_t size);
void librdf_sign_free(void *ptr);
They just use an extra int to track allocations so shouldn't present a
huge overhead for our usage needs. Though we get the librdf_sign_free
entry point out of the deal which to my understanding should allow
mitigation of the CRT issue, if that proves to be the issue.
This archive was generated by hypermail 2.1.8 : Thu Jul 14 2011 - 02:25:04 CEST