Craig, you are almost right :-). You can store annotations in the bytecode (check the class, field, method attributes), and we've done this in AspectWerkz and AspectJ for quite a while ;-). The problem is consuming them.
./alex -- .w( the_mindstorm )p. On 4/25/06, Craig McClanahan <[EMAIL PROTECTED]> wrote: > On 4/24/06, Don Brown <[EMAIL PROTECTED]> wrote: > > > > So... is that a +1? :) > > > > As for annotations, I'd imagine we'd keep the XML as override so most > > anything you can do with annotations, you could > > also do in XML. This would give Java 1.4 users mostly the same > > capabilities. > > > I think he was expressing a technical concern :-). And, depending on how > you want to use annotations, it's an important one. > > It is straightforward to see how one could use annotations at build time, > even if the ultimate target platform is 1.4. You can do things like run the > "apt" tools, generate new code (or maybe generate config files that > correspond to what the annotations signify) as part of the build process -- > as long as the build process itself was running on Java SE 5, of course. > Even this might be too much of a leap for some organizations, however. > > But the annotations facility also provides *runtime* mechanisms to examine a > Class instance and determine what annotations are present. I do not see how > that can be done with something like Retroweaver, for two reasons: > > * The annotations themselves can't be stored in the compiled class, because > there is nowhere to put them. > > * The APIs to do the reflective checks (added to the Class class in 1.5) > don't exist in the 1.4 runtime. > > This would be a pretty serious limitation on what you might otherwise want > to do with annotations -- but, again, it depends on the intended usage > patterns. > > Because of the above issues, I'm -0 on this, (it's not reasonable for me to > vote -1 since I'm not the one that is wanting to do the work). > > I think a better strategy would be to have the core framework depend solely > on 1.4, but provide an optional layer on top which leverages 1.5 things for > those who actually do have a 1.5 runtime environment. That's the approach > taken by the "Tiger Extensions" to Shale[1]. And, the first two features of > these extensions (use annotations instead of configuration files, and use > annotations instead of implementing interfaces) are *very* similar to the > sorts of things that would make 1.5 attractive in [action2]. > > Don > > > Craig > > [1] http://struts.apache.org/struts-shale/features-tiger-extensions.html > > Alexandru Popescu wrote: > > > This looks quite nice. I cannot figure out what means support for > > > annotations. Even if they are left inside the classbytecode, you will > > > not have access to the API to use them (except the case they are > > > weaving the Class.class, but I really don't think so). > > > > > > ./alex > > > -- > > > .w( the_mindstorm )p. > > > > > > > > > On 4/24/06, Don Brown <[EMAIL PROTECTED]> wrote: > > >> There has been a lot of discussion on Java 5 support for Struts Action > > 2, and from my reading of the comments, we have > > >> settled on a path, but I want to formalize it in a vote to ensure we > > are all on the same page. > > >> > > >> I vote we develop Struts Action 2 with Java 5, taking advantage of it > > where ever we can. At the same time, we should > > >> use Retroweaver to build jars that will run in a 1.4 JVM. For those > > that aren't familiar, Retroweaver supports > > >> conversion of an impressive amount of Java 5 features and language > > changes. In summary, Retroweaver supports [1]: > > >> > > >> * generics > > >> * extended for loops > > >> * static imports > > >> * autoboxing/unboxing > > >> * varargs > > >> * enumerations > > >> * annotations > > >> > > >> Therefore, our development philosophy will be to take _full_ advantage > > of Java 5, but provide a working jar for Java > > >> 1.4, however, we can't guarantee every Struts Action 2.0 feature will > > be available to Java 1.4 users. > > >> > > >> ------------------------------------------------------------------ > > >> [ ] +1 Make Java 5 the target > > >> [ ] +0 I am fine with this move, but I'll still mainly interested in > > 1.4 > > >> [ ] -0 I am not too keen, because ... > > >> [ ] -1 I am against this move, because ... > > >> ------------------------------------------------------------------ > > >> > > >> I'll tally the votes after at least 72 hours and include the count in > > our STATUS file. The vote is open to anyone. > > >> > > >> Don > > >> > > >> [1] http://retroweaver.sourceforge.net/documentation.html > > >> > > >> --------------------------------------------------------------------- > > >> To unsubscribe, e-mail: [EMAIL PROTECTED] > > >> For additional commands, e-mail: [EMAIL PROTECTED] > > >> > > >> > > > > > > --------------------------------------------------------------------- > > > To unsubscribe, e-mail: [EMAIL PROTECTED] > > > For additional commands, e-mail: [EMAIL PROTECTED] > > > > > > > > > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: [EMAIL PROTECTED] > > For additional commands, e-mail: [EMAIL PROTECTED] > > > > > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
