[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Re: [aspectj-users] policy enforcement for parameterless constructor and certain field declaration
|
Hi,
ok, your first two pointcuts aren't quite right:
First one:
declare error: hasfield((@Injected *) *): "hit match1"; // try to
simply match 'match1' as an @Injected field
This is attempting to match when the type has a field whose type is
marked @Injected. As far as I can see, that isn't what you want, you
are interested in whether the field itself is annotated, not the type
of the field. That would be simply:
declare error: hasfield(@Injected * *): "hit match1"; // try to simply
match 'match1' as an @Injected field
Second one:
declare error: hasfield(@Injected (!(@Managed *) *): "hit match2";
// try to match 'match2' as injection of a not @Managed type
you are just missing a parentheses
declare error: hasfield(@Injected (!(@Managed *)) *): "hit match2";
// try to match 'match2' as injection of a not @Managed type
Third one, I think that is a bug related to compound type patterns.
if you remove the within piece (and the &&) it behaves a little
better. Feel free to raise a bugzilla for this problem.
cheers
Andy
On 26 May 2011 15:10, Kristof Jozsa <kristof.jozsa@xxxxxxxxx> wrote:
> Hi Andy,
>
> thanks for your detailed answer. I understand you're dealing with a
> limited set of resources so certainly I did not want to complain about
> the lack of docs :) Now as you say, I checked the second edition of
> AspectJ in Action I own and it mentions the hasmember features but in
> about 10 lines as an experimental feature.
>
> At last I think I've found the unit tests I you mentioned in the Tools
> CVS here: http://dev.eclipse.org/viewcvs/viewvc.cgi/org.aspectj/modules/tests/?root=Tools_Project
> and grepped around in them but unfortunately I still struggle with the
> correct syntaxes of hasmember.
>
> Here's a test of the two use cases what I'm trying to accomplish (I
> did turn on XhasMember both in ajdt and maven):
>
> public aspect HasMemberTest {
> @Target(ElementType.FIELD)
> @Retention(RetentionPolicy.RUNTIME)
> public static @interface Injected {
> }
>
> @Target(ElementType.TYPE)
> @Retention(RetentionPolicy.RUNTIME)
> public static @interface Managed {
> }
>
> public static class Sample {
> @Injected
> private SampleManaged match1;
>
> @Injected
> private String match2;
> }
>
> @Managed
> public static class SampleManaged {
> public SampleManaged(int foo) {}
> }
>
> declare error: hasfield((@Injected *) *): "hit match1"; // try to
> simply match 'match1' as an @Injected field
> declare error: hasfield(@Injected (!(@Managed *) *): "hit match2"; //
> try to match 'match2' as injection of a not @Managed type
> declare error: within(@Managed *) && !hasmethod(new()) : "no default
> constructor"; // try to match SampleManaged non-default constructor
> }
>
> where the first declare fails to match and the other two is not
> syntaxically correct in spite of I've spent a couple of hours now
> trying to get them right (ok, the second one looks a bit too funky for
> me as well).
>
> Is type pattern matching complete with annotation usage?
>
> thanks,
> K
>
>
> On Thu, May 26, 2011 at 6:06 PM, Andy Clement <andrew.clement@xxxxxxxxx> wrote:
>> Hi,
>>
>> On 26 May 2011 01:36, Kristof Jozsa <kristof.jozsa@xxxxxxxxx> wrote:
>>> Hmm I wasn't yet aware of this hasmember feature. I tried to google
>>> around and also checked my copy of Aspectj in Action but could not
>>> find much examples.
>>
>> I can't recall if it was covered, but if it was it would be in the
>> second edition (not the first) of AspectJ in action. Given that it is
>> still an 'optional' feature, I'm not sure if Ramnivas did cover it or
>> not.
>>
>>> Are there any usage details on hasfield/hasmethod somewhere? Also, is
>>> there a unit test suite which I could check for examples? The AspectJ
>>> source download doesn't seem to include any tests..
>>
>> There are numerous tests in AspectJ for this. The source for the
>> tests is not available in zip form though, it is only in the
>> repository. The source zips downloadable alongside the distribution
>> are only for what is in the distribution. Searching the source I see
>> about 30/40 tests.
>>
>> The feature is actually described in this bug:
>> https://bugs.eclipse.org/bugs/show_bug.cgi?id=86818 - which you'll
>> actually see is still open. It is still open because you have to
>> switch on this feature with -XhasMember, it still hasn't been made
>> 'default'. It isn't default because it still doesn't play nicely with
>> all the other language features (basically ITDs) - until it does it
>> will still be something to opt into.
>>
>> But it was only in 1.6.9 that I added the ability to use type patterns
>> with declare warning/error:
>> http://www.eclipse.org/aspectj/doc/released/README-169.html
>>
>> ---
>> It is now possible to use a type pattern with declare warning and
>> declare error. For example:
>> declare warning: I+ && !hasfield(int i): "Implementations of I are
>> expected to have a int field called i";
>> ---
>> I do acknowledge (and apologise that) the main docs don't cover it
>> properly. We last had the resource to do a good docs update for 1.5.0
>> and that was all focussed on generics/annotations/etc when AspectJ was
>> upgraded for Java5. Since then it is basically the sequence of
>> READMEs that detail what has happened with AspectJ.
>>
>> Here is one of our testcases for it:
>> ===
>> public aspect HasMethod {
>>
>> declare parents : hasmethod(* print(..)) implements Printable;
>>
>> public static void main(String[] args) {
>> C c = new C();
>> if (! (c instanceof Printable)) {
>> throw new RuntimeException("declare parents : hasmethod failed");
>> }
>> }
>> }
>>
>> class C {
>> public void print() {}
>> }
>>
>> interface Printable {};
>> ===
>>
>> cheers
>> Andy
>>
>>> On Wed, May 25, 2011 at 3:16 PM, Matthew Adams <matthew@xxxxxxxxxxxxxxx> wrote:
>>>> For the first one, you might try using "declare error" with a type
>>>> pattern that uses hasmember. Not sure what you mean by the second,
>>>> but declare error with hasmember might do the trick there, too.
>>>>
>>>> -matthew
>>>>
>>>> On Sat, May 21, 2011 at 5:57 PM, Kristof Jozsa <kristof.jozsa@xxxxxxxxx> wrote:
>>>>> I'd like to ask for help with two policy enforcement rules:
>>>>> - enforce a class having no constructor with arguments (a default or a
>>>>> parameterless constructor is fine)
>>>>> - enforce an annotated field being of a certain type
>>>>>
>>>>> For the second one I currently enforce the get() of the field which is not
>>>>> too bad, but would be nicer to disallow the field declaration.. if possible.
>>>>>
>>>>> Can anyone show me an example doing these tricks?
>>>>>
>>>>> Thanks a lot,
>>>>> K
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> aspectj-users mailing list
>>>>> aspectj-users@xxxxxxxxxxx
>>>>> https://dev.eclipse.org/mailman/listinfo/aspectj-users
>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> @matthewadams12
>>>> mailto:matthew@xxxxxxxxxxxxxxx
>>>> skype:matthewadams12
>>>> yahoo:matthewadams
>>>> aol:matthewadams12
>>>> google-talk:matthewadams12@xxxxxxxxx
>>>> msn:matthew@xxxxxxxxxxxxxxx
>>>> http://matthewadams.me
>>>> http://www.linkedin.com/in/matthewadams
>>>> _______________________________________________
>>>> aspectj-users mailing list
>>>> aspectj-users@xxxxxxxxxxx
>>>> https://dev.eclipse.org/mailman/listinfo/aspectj-users
>>>>
>>> _______________________________________________
>>> aspectj-users mailing list
>>> aspectj-users@xxxxxxxxxxx
>>> https://dev.eclipse.org/mailman/listinfo/aspectj-users
>>>
>> _______________________________________________
>> aspectj-users mailing list
>> aspectj-users@xxxxxxxxxxx
>> https://dev.eclipse.org/mailman/listinfo/aspectj-users
>>
> _______________________________________________
> aspectj-users mailing list
> aspectj-users@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/aspectj-users
>
