(reposted to mason-users in hopes of getting more comments) We have a great deal of poorly-tested Mason code.
I've recently been using Catalyst some, and grown jealous of the testing tools available. Specifically, Catalyst::Test and Test::WWW::Mechanize::Catalyst. Catalyst::Test takes the name of a Catalyst application and gives you a simple API to make in-process requests (HTTP::Request and HTTP::Response). Test::WWW::Mechanize::Catalyst wraps that with a Test::WWW::Mech subclass. Both of them fake requests to http://localhost/, but can run the exact same tests against a real server based on an environment variable. This is much better than the solutions I had encountered before, which were things like using WWW::Mech on a development server or maybe on a temporary testing webserver instance (HTTP::Server::Simple). Devel::Cover works, tests are less complicated, flexibility for both local and remote, etc. There are probably some people for whom this is old hat, but it was new to me. I talked to Dave Rolsky about this at YAPC and he said that he didn't know of anything similar for Mason, so I'm thinking of trying to write it. A good deal of the work for this has already been done. I wrote a module (unreleased) that further abstracts the concept of a mason "site" (lots of configuration from YAML) and works with either HTTP::Server::Simple::Mason (well, technically any CGI) or Apache, and I'm sure I could pick MasonX::WebApp's brains too (or give it some of mine). The problems are mostly ones of interface and convention. 1) Catalyst has a clear idea of what an 'application' is. Does Mason? I'm not sure of the common usage, just our internal (and potentially pathological) one, where we were still using a handler.pl up until a few months ago :( This affects how to invoke the test module -- Catalyst does 'use Catalyst::Test "My::App"', and something reasonably close to as simple would be nice. 2) Even if Mason has a common usage, does it have one that can use either ApacheHandler or CGIHandler as necessary depending on the context? This might be a big problem, because it's much easier to get people to use a testing module they don't have to change their other code to accommodate. 3) Mason has a lot more things that you might want to potentially override for testing, especially when it comes to components. Catalyst effectively has the concept of a single comp_root, while when using Mason you may have 5 of them, 3 of which change (in different ways) for testing. 4) I'm not sure how well switching back and forth between http://localhost/ and a remote server will work in terms of URL handling. I can't think of a specific problem, but it seems like an area where gotchas could easily crop up. 5) ??? Other stuff I haven't thought of yet. Would anyone else be interested in this sort of thing, even if some changes to existing code were required in order to use it? Alternately, is there something that fills this niche already that I'm not aware of? hdp. Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Mason-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/mason-users
