(for review)  A long time ago, I wrote a very schlocky test called i18n/label.t.  It actually parses the WebGUI source code, looking for i18n calls that lookup table entries that don't exist.  It's very slow, taking up to 20 minutes even on very fast machines, but it has also found a lot of broken labels.

In january of 2007, our collective coding proved too powerful for my puny, pathetic parser.  So I began writing a Perl::Critic policy to do the same job.  As of June, it did a more thorough job of parsing than label.t did.  However, it runs about 3-5X slower than label.t.  Since it only broke in 1 place, I didn't think too much about it, and got used to seeing label.t failures in every nightly smoke run.

The new rewrite of the EMS breaks label.t pretty thoroughly, and exposes a bug in Text::Balanced, which hasn't seen a new release since December of 2006.

I think it's time to replace label.t with the Perl::Critic policy.

However, that 3-5X slow down (that's 60-100 minutes!) makes me pause so I figured I'd better check with my fellow devs.

What do y'all think about replacing the broken test with the slow but functional test?

Should that be better named? 

Something like: DEV_ONLY works for me.

I agree with slow and correct, by the way. So long as it doesn't  
affect the loading of WebGUI as users see it, I don't see a problem  
with taking a bit of time to do something important.

The WebGUI tests don't interfere with loading WebGUI, so we should be safe there.