https://issues.apache.org/bugzilla/show_bug.cgi?id=56399
--- Comment #5 from Konstantin Kolinko <knst.koli...@gmail.com> --- In reply to Mark's e-mail on "Re: r1594436" 2014-05-14 12:02 GMT+04:00 Mark Thomas: > Why is an exception being used for flow control? Surely a boolean flag > would be better here especially as this is being called on every > request. Ack. In some sense those are remnants of older code that I had before simplifying the bits. I do not think those slow down things (as throwing the exceptions is an unexpected use case). I still think that exceptions are needed for the log statements, but we can generate the exceptions only when the log messages are actually written out. As I mentioned in comment 4, it would be better with separate message texts for unrecycled request vs. unrecycled response. (BTW, The mailing list still fails as a place for discussions. This particular message was delayed by 48 hours). 2014-05-14 12:02 GMT+04:00 Mark Thomas: > I remain of the view that a better solution would be to recycle both > pairs of request and response objects at the same time. The idea of "to create a new Recyclable interface in Coyote" (from Comment 0) is not doable. The "notes" in coyote request are opaque. The CoyoteAdapter.ADAPTER_NOTES constant is defined in Catalina only. (In coyote it is mentioned in javadoc for coyote.Request#setNote()). Thus a better solutions looks to introduce some method into Adapter interface. (Or rename and adapt the "checkRecycled()" or "log()" one). Will the recycling driven by Coyote or by Catalina? Will some code from Catalina be removed or simplified? Will the need to recycle accompanied by some action? Will it be possible to backport the changes to earlier versions? -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
