Amanda Waite wrote: > Nick Kew wrote: >> Strong +1 in principle. I hadn't even realised these were separate >> from mod_fcgid as already packaged (OK, I wasn't looking)! >> > > Cool, good to know that it's worth doing. >> It's a very stable interface: CGI and mod_fastcgi haven't moved >> either (though the latter could be because mod_fcgid more-or-less >> obsoleted mod_fastcgi). >> > > I agree, but if something important breaks then dependent on > circumstance we'll probably need to find a way to fix it. > >> >>> - Is it really necessary to include 32-bit and 64-bit versions of >>> the executable? >>> >> >> Is there any reason not to? >> > > No, at least until an Arc member asks why we included it.... > >> >>> There also doesn't appear to be any bundled tests which is a bit of >>> a problem. >>> >> >> Interesting. Do you have anything in mind as a basis for a test kit? >> I'm just wondering whether it might be worth revisiting cg-eye and >> building a few tests on top of it. >> > > It needs some thought I guess, any test 'client' would need to be > FastCGI savvy and generate requests and understand responses. I'd need > to make a list of what the functionality was and then decide on what > functional tests would look like, work out how best to implement them. > I'll look at the CGI angle. Do you know of any tests for mod_fcgid/fcgi? In lieu of any real test suites:
http://blogs.sun.com/trawick/date/20080820 (if those instructions do not get any easier, we have done something wrong) Apache2 + mod_fcgid + Ruby + lib-FastCGI testing could be done with a simple C-language FastCGI app as well as a simple Ruby (fcgi) FastCGI app. Hopefully (duck) there is already some sort of extendable test framework for OpenSolaris web stack that can start the server with some configuration and issue requests and check the results?
