09.05.2011 3:50, Konstantin Kolinko пишет:
2011/5/7 Fedd Kraft<f...@sql.ru>:
Hi,
I think I have problems in my software because the implementation of EL
works so that the set value is coerced to the target type.
I am talking about lines 153-157 of the class org.apache.el.parser
<http://www.docjar.com/docs/api/org/apache/el/parser/package-index.html>.AstValue.
I guess it makes sence all the time except if the 'base' is the Map. I seem
that once I put a String into the Map, I lose the ability to put, say, an
array or a Collection there through the setValue method of the
ValueExpression.
The ValueExpressioni s a property of the Map, like this ${map.key}. If this
expression is String, I can't setValue a Collection there, my collection
becomes a String through toString method, because the resolver.getType
method returns the Class of what is currently kept under this key.
Please tell me whether this is an intended/unavoidable behaviour, or a bug?
I think you can file a bug, but please provide a working example that
reproduces the issue.
If I see it correctly, reading "expression_language-2_2-mrel-spec.pdf"
1) ch.1.2.1.1 says
"In the case of lvalues, the expected type is ignored and the provided value is
coerced to the actual type of the property the expression points to, before that
property is set."
2) assigning of a value to a map entry should be handled by
MapELResolver, which Javadoc is in ch.2.15. The getType() call there
is documented to return Object.class.
Implementation of javax.el.ELResolver#getType() in trunk performs a
bit differently.
BTW, a workaround in your case might be to do assignment in two steps:
first assign null, and then assign the actual value.
Best regards,
Konstantin Kolinko
Konstantin,
I'll try to prepare a reproducing example a bit later if I can...
unfortunately I can't even set the value of an expression (with setValue
method) to null, as null gets coerced to an empty String.
Is far as I recall, approximately two years ago it worked as I expected,
then I guess the map resolver was changed to return a type of the value
kept instead of Object.class, and it was made for some reason.
Anymay, thanks, Konstantin, I'll try it some other way.
(happy Victory Day BTW :)
-fedd
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org
