- a way to avoid seeing intra-package dependencies between classes since classes in the same package are supposed to be releated anyway and in fact, there's no reliable way to spot all of them anyway (since you don't need imports for using classes in the same package)
- a way to limit the visibility of members used to perform dependency inference.
Both of these allow to reduce the "fur ball" effect.
Finally, removed the -version2 option since we can rely on the -quiet standard javadoc option if we use RootDoc.printNotice/Warning/Error
All in all, having fur balls generation is not that bad, in a sense it makes people see what classes/package are brittle and which are on the contrary crisp, better defined...
* output using the RootDoc.notice/warning methods instead of system.out (standard way, and allows the output to be seen from maven2 javadoc plugins as well)
* works also if the subpackages parameter is not specified on the command line
* added javadoc comments
* other smallish fixes
Also, add an option to avoid overriding class hiding specified in the parent option provider, since the hides specifed for the doclet should not be modified in a context class diagram
* ability to compute relative links for UmlDoc diagrams (client side image maps)
* better handling of sources with annotations and generics usage
* avoid relying on specifiedpacakges to check if classes are included in the parsed source, maven 2 does not provide javadoc with them, but with a plain list of classes
Also added a reflective toString() that I've used for debugging settings during UmlDoc developtment (the code comes more or less from the first pages of Code Quality, the open source perspective :-) )
Diomidis, can you have a look at the output (UmlGraph javadoc are built with UmlDoc now) and review the code?
I know I still have to write the docs, and I will have to try out UmlDoc on something other than UmlGraph itself (was thinking of the java sources themselves, or Hibernate, Spring, ... whatever is big enough).
un-noticed since I never wiped out the testdata-out folder contents. Anyway,
I've fixed the bug and made sure all .dot and .png files are removed from the
folder as the test starts, so that a similar problem won't stay hidden in future.
Anyway, the error you sent me seem to be something different (possibly HTML
code generation, dot is very picky about what HTML it can parse).
Can you try again with this patch and let me know? If you see again that error,
can you send me a few of those failing dot files?
Moreover, what dot version do you have? I have version 2.6 (Sun Aug 28
13:14:52 UTC 2005).
About my aliceposta mail address, I don't know, I'm receiving mail so it cannot
be fully broken, but yes, it's not the first message I don't receive this week.
The mail you sent me thru istlab.dmst.aueb.gr arrived fine (but it's just a
redirector, isn't it?).
Wow... pretty long for a commit message :-)
I'll be online again in the late afternoon. See you.
Andrea
Andrea Aime