You are of course, correct. I was thinking too local in scope and
totally overlooked developers deploying to different environments.
As for the multiple templates, I guess I misunderstood what you were
initially proposing. Dealing with it at time of class generation
would be a much cleaner method.
Re: EOL property, we should still have it set, but for reasons
completely tangential to the issue at hand now. The ASF provides
default svn config file to auto add the EOL property for any new
files, too.
--
Kevin
On Jun 2, 2008, at 8:17 AM, Andrus Adamchik wrote:
On Jun 2, 2008, at 3:04 PM, Kevin Menard (JIRA) wrote:
Correct. But the Velocity templates in the Windows artifacts can
use Windows EOL, the Mac artifacts Mac EOL, and the tarballs Linux
EOL.
I am strongly -1 on it. This assumes that Cayenne runtimes jars from
our "windows" download will be used on Windows - an unwarranted
assumption (as you mentioned yourself later in the comment).
This would obviate the need for maintaining multiple versions of
the Velocity templates that only differ by EOL.
I am not advocating having multiple templates. If we want platform
line endings, we should do something smarter, like intercepting that
at the Writer used by template or similar.
Granted, this isn't as foolproof as a runtime OS check -- e.g.,
there's nothing stopping me from using the tarball on Windows and
getting the wrong EOL. I suspect this isn't a case that arises
frequently, except when a user grabs an artifact from the maven repo.
It arises all the time. The Maven example is an excellent one, that
I haven't thought about. But more generally any large group of
developers, corporate or open source, maven or jars-in-svn, will
have to deal with it (e.g. develop on Windows, build/deploy on
Linux, but sometimes on Windows too). I would say cross-platform is
not an exception - it is the rule.
Andrus
