Not actually a ruby related question, but does anybody know the
equivalent of auto test for a maven based java project?
Thanks!
-r
On Fri, Nov 6, 2009 at 10:40 AM, Roger P. [email protected]
wrote:
Not actually a ruby related question, but does anybody know the
equivalent of auto test for a maven based java project?
Thanks!
-r
I actually don’t know of any, and when I found autotest in Ruby I
always wondered why nothing similar had been done for Java. You might
try using a Ruby library like directory_watcher and just have it
re-run maven when files change.
/Nick
To unsubscribe from this list, please visit:
http://xircles.codehaus.org/manage_email
On Nov 6, 2009, at Nov 6,8:55 AM , Nick S. wrote:
I actually don’t know of any, and when I found autotest in Ruby I
always wondered why nothing similar had been done for Java. You might
try using a Ruby library like directory_watcher and just have it
re-run maven when files change.
There’s also the watchr gem. A number of watcher-like gems … too
many, at least from the perspective of trying to build momentum.
I’ve been thinking about the same issues re: integration with the
build tools. autotest was kind of a revelation when it came to ease of
testing. But that’s just the beginning. What about building HTML from
Haml or CSS from Sass (just a couple of my itches). Or compiling .o’s
or .class’s. Basically, everything our build tools do, whether they’re
maven, or ant, or rake, or make. Why shouldn’t those be more event
driven? Coming from autotest, the batch mode feels archaic and annoying.
I suppose the IDEs have mitigated some of these issues, for
environments/users where they’re used. But lots of non-IDE things
naturally fit this, too.
But, of course, those build tools aren’t structured like that. I find
it a very interesting idea.
In my own experiments, when you do start down this path, things get a
little tricky, quickly, with regard to dependences. Even autotest
doesn’t do a particularly good job at this. The problem isn’t
implementing dependences, but discovering them. gcc was great with, -M
was it? And I had a whole environment that used jikes just for -M. But
it seems like the cases where dependences are detected and
externalized are the very rare exception. Hierarchical/multiple-pass
makes are equally ubiquitous and evil.
Ah, well, enough of my soap-box. Sorry.
To unsubscribe from this list, please visit:
http://xircles.codehaus.org/manage_email