[CODEC-25] [Codec] Default URL encoding logic broken

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.2
• Fix Version/s: 1.3

Environment

Operating System: All
Platform: All

• Bugzilla Id: 25995
• Assignee: Unassigned
• Reporter: Oleg Kalnichevski
• Created: 09/Jan/04 02:03
• Updated: 27/Oct/07 06:49
• Resolved: 27/Oct/07 06:49

Description

It looks like there are some changes made to the URLCodec class(rev 1.7 and 1.8)
that do not seem quite right (in my opinion).

http://cvs.apache.org/viewcvs.cgi/jakarta-commons/codec/src/java/org/apache/commons/codec/net/URLCodec.java?r1=1.7&r2=1.9

The ENCODING variable in my original patch was made final static for a good
reason. It is not supposed to be anything but US-ASCII. After all, the whole
point of URL encoding is to represent non-ASCII data using a limited set of
ASCII characters only, hence, 'US-ASCII' as a final static variable ENCODING.
With the present code, if US-ASCII incompatible encoding is used (such as
UTF-16, EBCDIC), URLCodec may produce erroneous results. I understand that the
intention of the revision 1.7 & 1.8 was to allow for user defined default
encoding, but in its present form URLCodec does not seem to produce the desired
results.

The patch I am about to attach should take care of the problem while retaining
the existing (intended) semantics of the URLCodec class.

Oleg