Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to email@example.com.
"Tony Hoyle" <tony.hoyle at march-hare.com> wrote in message news:dlghti$t2m$2 at paris.nodomain.org... > Jan Giesen wrote: > > > Under Linux I get the info "checkout trigger not enabled", but not under > > Unix. > > Both executables are compiled from the same source code and operate on the > > same > > repository files. This behaviour is seen on 2.5.02.2115 as well as under > > 2.5.02.2151. > > What could be the reason for this difference?? > > The 'not enabled' comes from the trigger itself, depending on the > setting of 'CheckoutTrigger' in /etc/cvsnt/Plugins. > > In the HPUX case it seems that this plugin is enabled. > > Tony An investigation of the problem has shown that under HPUX the loading of the trigger libraries is broken: Only the first trigger library "default_trigger.sl" (= info_trigger.cpp) is loaded correctly. The corresponding "init" function is called correctly. But loading of the next trigger libraries e.g. audit_trigger.sl goes wrong: The corresponding GetCvsInfo function is still invoked and returns a pointer to a "callbacks" struct. But the entries of the callbacks struct are still initialized with the values of the first trigger library "default_trigger.sl". Therefore the "init" function of the default_trigger lib is called instead of the "init" function of the "audit_trigger" lib. This is the reason, why the info "..trigger not enabled" is not generated. When I rename "audit_trigger.sl" to "default_trigger.sl" indeed the "init" function of the audit_trigger lib is called. What could be the reason that the "callbacks" struct is not correctly initialized?? Best regards Jan Giesen