Remy Maucherat wrote:
On Fri, 2008-03-21 at 11:25 -0600, Filip Hanik - Dev Lists wrote:
what this filter does, is it relies on the already implemented
BufferedReader, no need to reinvent the wheel.
The java.io.BufferedReader, already has this logic implemented, and it
works, so CoyoteReader, will only be treated as a java.io.Reader, and we
can bypass all the stuff in Tomcat's code base
Maybe this piece of code is hopelessly broken and should be removed. If
it's not, I think some effort should be made towards improving it [as
well as improving the character conversion code], as double buffering is
not something nice and a BufferedReader is a relatively expensive
object.
I'd prefer to move forward with just the truncation fix so that we can
move towards a release.
Any other efforts can be postponed at this time. If the mark/reset bug
in fact exists, there is an easy work around, and that is to use
java.io.BufferedReader.
whether the code is hopelessly broken or not, I don't have enough info
to comment on. But I do think the code is too fragile to be mucked with
in our 6.0.x branch, that can be done in trunk or sandbox.
It's better for us to get a release out that takes care of the
regressions that are in the releases we have now.
Filip
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
