Details
-
Improvement
-
Status: Closed
-
Major
-
Resolution: Fixed
-
current (nightly)
-
None
Description
The axis code concatenates its own message text with the text from the HTTP messages . For example, if there is a basic authentication failure, it concatenates "HTTP transport error" with "unauthorized". So, the caller sees the text "HTTP transport error unauthorized". There is no way for the caller to know what the HTTP message was withouth parsing the exception text. The caller can take appropriate action based on the HTTP message. If the HTTP message says "unauthorized" the operation can be aborted. Other wise (server down or some network glitch) the operation can be retried. So, it would be nice if the axis code returns its message text and the HTTP text separately. Maybe there should two "what()" methods, one returning the axis text and the other the HTTP message text.
This problem might be there for other types of exceptions also.
Sounds like we need some kind of "cause" method that returns the wrappered problem?