Question

So, I’ve spent enough time using ASP.NET webforms to know that I’d almost rather go back to doing classic ASP than use them. But I’m hesitant to move to ASP.NET MVC until it becomes more mature. Are there any open source alternatives?

The main thing I’m looking for is something that’s easy to learn and to get a prototype up and running with. If it’s any help, my main preference is a python ‘mix and match’ approach (like say web.py/SQLAlchemy/whatever templating engine suits my fancy at the time).

Answer 1

We have used MonoRail RC2 for our small business’s online store for the past 18 months. It replaced a 7 year old disaster of classic ASP pages. MonoRail RC2 has worked well for us, serving an average of ~14,000 page requests per day. It enabled me to develop the site very quickly, was free, and holds up well. For that, I am thankful to the MonoRail team.

I just used the MonoRail bit. I chose iBATIS.NET over ActiveRecord as I had to write creative SQL to maintain compatibility with a 7-year-old turd of a database. So I can’t speak for some of the other Castle libraries.

Some advantages of MonoRail include the following:

• It’s pretty easy to get in there and modify things. For example, the default routing implementation didn’t preserve the query string when rerouting (I needed this to preserve backwards compatibility with the old URL formats used by the now defunct classic ASP site for SEO reasons) and it didn’t support issuing HTTP 301 Permanent Redirect headers for that scenario. So I implemented whatever the MonoRail interface for that was, plugged it into my config file, and off I went.
• It’s still ASP.NET, so you can still use Forms Authentication, the totallly awesome HTTP/HTTPS Switcher at Codeproject, and caching.
• Relatively few surprises. After working with it for over year, I haven’t had too many days where I’ve cursed at MonoRail. That’s a pretty good litmus test. Some of the helper classes (FormHelper) can behave a little strangely, the wizard framework is outright bizarre, and parameter binding can sometimes throw you for a loop, but it doesn’t happen often.
• Choice of View Engines (templates). I put this here because most people seem to think choice here is a good thing although I usually think that it is not.

MonoRail, however, is not without its problems:

• Lack of direction in development. The number of changes between RC2 and RC3 was beyond ridiculous; a lot of protected virtual methods disappeared, a lot of helpers changed (which is a big deal when your view engine isn’t statically typed), even the mechanism for unit testing controllers and views changed. For this reason, we’re probably just going to stay on RC2 forever. Now that ASP.NET MVC is out, it’s unclear how healthy the community behind MonoRail will stay (though ayende and hammett are as enthusiastic and active as ever).
• NVelocity, the ‘de facto’ view engine for MonoRail (at least at the time that we started development), is a promising templating language with an unhealthy implementation and maintenance outlook. (Does it work? well enough. But being a CTRL+C CTRL+V port from the Java version, don’t read the source for that library because your eyes will bleed.)
• NVelocity and RC2 shipped with an extremely severe threading bug where multiple users accessing the site at the same time could get served pages intended for the other. It’s fixed in the latest release (which, due to the release nature of the Castle projects, is very difficult to upgrade to), and we managed to work around it. But it was a very disturbing and unexpected issue to encounter, one that would be highly unlikely to be encountered on a Microsoft framework. Caveat emptor.

MonoRail provided an excellent opportunity for us in June 2007 by giving us a way to migrate an existing site on the Microsoft stack to the .NET platform in a way that avoided WebForms (which is great for intranet sites, but not so great when you need fine-grained control of your HTML output on a public-facing Web site, in my opinion). (Fine, the real reason is I just despise the WebForms postback model.) ASP.NET MVC was not even a gleam in Microsoft’s eye at that point in time.

Now that ASP.NET MVC does exist, however, and given that Microsoft is positioning it as an alternative to WebForms, I know that I personally will strongly consider it for any future project. MonoRail is a great project, it served us well, and I’m grateful to the open source community for it, but I think of it fondly as a heavily used, worn tool that is retired to a lower drawer in my workbench. Without it, ASP.NET MVC might not exist today.