David Golden wrote:
On Jan 29, 2008 12:56 AM, Michael G Schwern <[EMAIL PROTECTED]> wrote:
2) The most common mistake is forgetting about dependent actions when you do
a complete override, and that's not documented anywhere.
Sure it's documented -- it's documented in the code. I can't imagine
doing a complete override of a function and not actually looking to
see what happens in that function that I'm replacing.
Some dependencies are more complicated, like testcover depends on code, then
does some work, then "depends on" test when in reality it doesn't really
depend on the test action but overrides its behavior. testdb is similar.
Either way, complex cases like this can still be handled, depends_for() is a
convenience, not a replacement for depends_on().
Should depends_on change the list of dependencies?
There should be some sort of mechanism I guess, but probably not. That gets
into depends_on() figuring out what action it's currently in which gets into
looking at the call stack and oh god messy.
More practically, I don't think calling a dependency should be tied to
declaring a dependency. If the mechanism were more formalized and
declarative, like in make, the yes. But if they're tied together with the
actual running of the action, then no.
Speaking of declarative dependencies, making them more declarative would help
to enable parallel builds.
--
94. Crucifixes do not ward off officers, and I should not test that.
-- The 213 Things Skippy Is No Longer Allowed To Do In The U.S. Army
http://skippyslist.com/?page_id=3
