On Sat, Apr 26, 2008 at 2:03 PM, simon <[EMAIL PROTECTED]> wrote:

>
> On Fri, 2008-04-25 at 19:39 -0500, Leonardo Uribe wrote:
> >
> > 2008/4/25 Andrew Robinson <[EMAIL PROTECTED]>:
> >         Forgot to ask, this uses annotations for JSF >= 1.2 right?
>
> Yes.
>
> Well, actually it is user's choice, as both are supported. But I would
> imagine that people would choose real annotations when available.
>
> >
> >         Also one more question, would Jsf* be better than JSF* for the
> >         names
> >         (ie JsfProperty vs JSFProperty)? Using all uppercase for an
> >         acronym in
> >         a java name is not conventional. Uppercase is usually reserved
> >         for
> >         starting a new, unabbreviated word.
> >
> > Yes, I have seen this too. The convention on xdoclet is use
> > namespace.prefix. I would like to use something like this:
> >
> > oam.Component
> > oam.Converter
> > oam.Validator
> > oam.RenderKit
> > oam.Renderer
> > oam.Property
> > oam.JspProperty
> > oam.JspTag
> > oam.JspAttribute
> > oam.Exclude
> >
> > or something like that. I would like to see what Simon thinks about
> > this, since he proposed the names at start and I don't know if he has
> > a particular reason about the choice of the tags.
> >
> > Simon: What do you think about this?
>
> I would suggest that it is nicer to have the javadoc tags and real
> annotations use the same names. A class obviously cannot have "oam." in
> the name. Possibly the real annotations could have the above names with
> the "oam." stripped off. However in at least some cases, the result
> seems a too-generic name to me. For example, @Property or @Component
> aren't immediately obvious when reading the code, and may clash with
> other libs (doesn't Seam have @Component already?).
>
> I think in Eclipse, using a dot name for a javadoc tag also makes the
> spelling-checker mark the name as a spelling mistake, which is mildly
> annoying.
>
> If people want JsfComponent rather than JSFComponent, I'm ok with that.
> But Sun went to the effort of actually deprecating getUrl() in favour of
> getURL(), so presumably the all-caps-for-acronyms is a java convention.
>
>
Looking the java api there is a lot of classes that when use an acronym at
start,
it put all upercase like this classes:

RMIConnector<file:///C:/jdk-1_5_0-doc/docs/api/javax/management/remote/rmi/RMIConnector.html>
RMISecurityException<file:///C:/jdk-1_5_0-doc/docs/api/java/rmi/RMISecurityException.html>

AWTEventMulticaster<file:///C:/jdk-1_5_0-doc/docs/api/java/awt/AWTEventMulticaster.html>
POAHelper<file:///C:/jdk-1_5_0-doc/docs/api/org/omg/PortableServer/POAHelper.html>

SQLException <file:///C:/jdk-1_5_0-doc/docs/api/java/sql/SQLException.html>
URLEncoder <file:///C:/jdk-1_5_0-doc/docs/api/java/net/URLEncoder.html>

So if we want that Doclets looks like annotations, the best is let the
convention as is.

Thanks Simon for make clear this point.

regards

Leonardo Uribe


>
> >
> > On myfaces 1.2 maybe annotations could not be used, because maybe it
> > breaks the TCK, but I'm not tested this yet, so its a point to take
> > into account. The actual code allows to use annotations and doclets
> > (we need to update annotation classes).
>
> I think it *very* unlikely that annotations would conflict with the TCK.
> These are source-retention annotations only, ie have absolutely no
> effect on the generated .class files.
>
> So if this plugin is used with myfaces 1.2.x, I would recommend real
> annotations be used and not doclet tags.
>
> Note that using qdox to scan source for annotations is actually *easier*
> than using APT for it, for several reasons. The APT tool is really ugly
> to use. So there is no reason to think of the myfaces-builder-plugin as
> having a "legacy" architecture just because it also supports doclet
> tags.
>
> Regards,
> Simon
>
>
>

Reply via email to