From: David PONCE <david.ponce@wa...> - 2006-01-06 13:19:12
> I think that it is supposed to find partial names since it is a
> utility for other types of search, like word-searh, though I'm not
> positive. Hopefully David Ponce can provide some insight.
This is a known limitation of `senator-search-tag-name' which uses an
heuristic to locate the tag name in the buffer, as the tag doesn't
provide itself this kind of information (locations of tag parts like
the name, modifiers, etc.).
Please, read the thread "possible bugs in senator.el and
wisent-java.wy" in the cedet-devel ML archive at:
AFAIK, no good solution has been found to this problem for now.
From: David PONCE <david.ponce@wa...> - 2006-01-09 11:38:33
> While using JDEE to develop Java code in Emacs, I found a bug in
> senator-search-tag-name. Finding the definition of the following
> symbol from another class fails:
> FooBar foo;
> It fails in senator-search-tag-name, because it neglects to add
> "\\>" to the end of the regexp it uses. So instead of finding foo,
> it finds FooBar. Adding "\\>" to the regexp fixes it:
After a more thoroughly analysis of your fix, and how senator search
works, I think your fix was good. So I committed a slightly different
version of it into CVS.
I also modified the Java 1.5 grammar in wisent-java.wy to correctly
setup the tag bounds of compound variable declarations to fix senator
incremental search in some corner cases.
To compile the new grammar, type "make wy" into your semantic/wisent
directory. Provided you included the following in your .emacs file
after loading cedet.el to enable use of the Java 1.5 grammar:
;; Use the full Java 1.5 grammar to parse Java files
(autoload 'wisent-java-default-setup "wisent-java"
"Hook run to setup Semantic in `java-mode'." nil nil)
Thanks for your report.