mina-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "W.B. Garvelink" <barend.garvel...@gmail.com>
Subject Re: IoService API Growth
Date Sun, 13 Jul 2008 12:25:04 GMT
Attached is a first proposal, please take a look. I'd like to see if
there's any initial feedback now, and I'll work up a more extensive
patch the coming weeks before M3 is released.



Open issues:
- should maybe leave the methodes on the IoService interface for the
next milestone, marked deprecated, and don't remove them until 2.0
final.

- interface naming is open to debate (e.g. the current naming scheme
doesn't really allow the same interfaces to be used for IoSession
statistics).

- location of the lastReadTime() / lastWriteTime() / lastIoTime()
methods is open to debate.

- to keep the impact of this first patch low, I just made
AbstractIoService implement the statistics interfaces, rather than
factoring them into inner classes. (separation of concerns, high
cohesion, etc.) Factoring them into separate classes gives the ability
to deploy those instances into an MBeanServer, which I think is a big
reason to make that change. Not all application servers use the
standard MBeanServer (at least historically; Java 5 may have changed
that) and JMX+OSGi isn't always straightforward, so the actual MBean
deployment should be handled by a pluggable strategy implementation.

- I haven't touched the IoSession statistics yet, but the basic idea
is the same.

- anything else?


Thanks,

Barend



On Fri, Jun 27, 2008 at 7:22 AM, peter royal <proyal@apache.org> wrote:
> i think that's a very reasonable improvement. care to work up a patch? :)
>
> -pete
>

Mime
  • Unnamed multipart/mixed (inline, None, 0 bytes)
View raw message