OK, I've got Checkstyle 2.1 now.  I was fooled by the fact that there was no
news article for it on the Checkstyle web site, so I thought 2.0.1 was the
latest.

As far as the @see goes, I think if you review the current code you'll find
a substantial number of occurrences with long lines.  I think it's a waste
of time to wrap these and having them flagged as errors will obscure other
more important errors, discouraging developers from fixing other more
important style violations.

I actually find it useful to be able to tell at a glance which class or
interface a particular method comes from, so I think we should allow (but
not require) the non-Javadoc @see comments.  Of course, Eclipse doesn't wrap
them when it inserts them.  Talking about short package and class names
isn't really relevant now since those were decided on long ago and are
basically fixed now.

The only disadvantage that I see to /* @see */ is that it has the potential
to get out of date during refactoring.  I think this is outweighed by the
advantage of having quick reference which is usable on paper or in an IDE
without hovering over the name.

What is the disadvantage to allowing them in the default format that Eclipse
uses to insert them (ie unwrapped)?

Tom

-----Original Message-----
From: Linus Tolke [mailto:[EMAIL PROTECTED] 
Sent: Tuesday, October 24, 2006 1:24 PM
To: [email protected]
Subject: RE: [argouml-dev] The new version of checkclipse


Hello Tom!
 
I have downloaded checklipse 2.1.0 and that has checkstyle 4.2 bundled. I
guess Michiel has the same because he suggested the logLoadErrors solution.
I didn't think more about it since it worked in my Eclipse.
 
The checklipse 2.1.0 has a problem in that the old .checklipse files are
replaced by a .settings/de.mvmsoft.checklipse.prefs file and the information
about files not to be checked is lost in the process. The problem with this
is that I cannot further update the .checklipse file. We will have to decide
when we shall stop supporting the .checklipse file format.
 
No longer allowing long @see lines is intentional. The reasoning behind it
is that I thought that their use was more or less only the required @see
comment of overriding and implementing methods. With that requirement gone,
the need for allowing this is reduced. Admittedly there is still some use
but I think that would be slight. Especially since the use of very long
package and class names is not recommended.
 
I added the flagging of non-Javadoc when I noticed that Eclipse by default
had the javadoc comment for new methods set to
/*
 * (non-Javadoc)
 */
I very much prefer no comment (in violation with the design guidelines) to
this comment (in violation with the design guidelines) added to every
method. Since I did this we have defined a template for methods and I think
that this or some other Eclipse maturity has reduced the problem.
 
In what cases would such a non-Javadoc comment be useful? My Eclipse colors
the javadoc comments and the non-javadoc comments differently so for me the
problem of distinguishing them is so small that I don't feel the need to
have non-Javadoc written in every such comment. I guess most non-Javadoc
comments use the //-syntax.
 
            /Linus

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to