> > On 12/08/2008, at 4:05 PM, Amos Jeffries wrote: > >>> >>> On 16/07/2008, at 4:12 PM, Benno Rice wrote: >>> >>>> Here are two patches against squid 2.HEAD. >>>> >>>> The first completely changes the nature of method_t in order to >>>> allow squid to pass through methods it doesn't know about. What is >>>> currently called method_t becomes method_code_t and method_t becomes >>>> a structure containing a method_code_t and a string for the method >>>> itself. We have a bunch of pre-allocated method_t's for methods we >>>> know about and any method we don't know about ends up with a code of >>>> METHOD_OTHER and a specially allocated method_t. >>>> >>>> The second patch builds on the first by adding proper purge >>>> behaviour in response to unknown methods and methods that are >>>> specified as causing object purges. >> >> By that do you mean you are syncing with the logic fix made for 3.1? > > I didn't directly refer to that, no. I set up a set of simple unit > tests that checked compliance against the RFC and tweaked it until it > passed.
Okay. Nice. The two versions will need to be checked for differences then. We don't want to loose any improvements done in the back-port. If the unit-test can be applied to 3.x too that would be great. Amos
