[reSIProcate] Updates to MethodHash and test drivers.
- From: Alan Hawrylyshen <alan@xxxxxxxxxx>
- Date: Wed, 23 Jun 2004 18:41:59 -0600
So it turns out that the autotools based build system was not building
the testHash program.
It also turns out that the testHash program didn't check the METHOD
hashes.
Also turns out that whomever added the UPDATE method didn't add it to
the name array.
This meant that under certain circumstances, using the UPDATE method
may have resulted in accessing memory beyond the end of an array.
I think this is why there were helper routines (getMethodName())
once-upon-a-time instead of using the MethodName[] array directly.
At any rate. There are other changes that I made to bring the Method
Types more in line with the HeaderTypes, namely I moved the definition
of
UNKNOWN from the END of the enum to the beginning, and set it's value
to -1. Just like the HeaderTypes.
If you recompile your applications and resiprocate together, you will
be fine. Otherwise you might have odd problems.
This would count as an ABI incompatible change -- so the next release
of the shared library will have a version increase.
r3028 contains these changes.
See the commit mailing list for diff links.
New Revision
3028
Committer
alan (Alan Hawrylyshen)
Date
2004-06-23 20:36:41 -0400 (Wed, 23 Jun 2004)
Log
Added METHOD hash test.
Fixed memory corruption problem if using the UPDATE method.
Unhooked gperf autogeneration.
Modified:
• main/sip/autotools/sip/Makefile.am
• main/sip/autotools/sip/resiprocate/test/Makefile.am
• main/sip/resiprocate/MethodTypes.cxx
• main/sip/resiprocate/MethodTypes.hxx
• main/sip/resiprocate/test/testHash.cxx
a l a n a t j a s o m i d o t c o m