Paul, I'll update the wiki tonight with your suggestions. You asked earlier about:
"Can the following options also be used on a STOMP transport connector? watchAdvisoryTopics, alwaysSessionAsync and dispatchAsync" *Transport connectors* are a broker configuration, ie, <transportConnector uri="..." /> The options you asked about, watchAdvisoryTopics, alwaysSessionAsync, etc are settings to make on the client side as the configure the client connection. transport.* options can be made on client or broker side as they configure the underlying transport objects. Sorry for the confusion. On Sun, May 12, 2013 at 2:29 PM, Paul Gale <paul.n.g...@gmail.com> wrote: > Hi Christian, > > I am still a little confused given your previous answer. At first you > state: > > >No. those configure the TCP transport. They are not openwire or STOMP > >exclusive. They are also *not client or broker specific*. I can update the > >wiki, but they can more explicitly be set as '*transport.soTimeout*'" > > Then later you say: > > >> transport.ioBufferSize=16384 > >> *transport.soTimeout=10000* > >Yes. But regarding your question previously asked... those options are > *reserved > for client-side settings.*" > > You've referred to the same setting (in this case transport.soTimeout) as > being '*not client or broker specific*' and then later '*reserved for > client-side settings*'. Can you clarify which it is? > > So with regard to these two pages in particular (and I daresay others): > > http://activemq.apache.org/tcp-transport-reference.html > http://activemq.apache.org/configuring-wire-formats.html > > the tables of definitions could benefit from an additional column > indicating the setting's applicability: client, broker or both. Thoughts? > Such a column would definitely help me out. > It would also help if the wiki more clearly specified what the option > prefix should be. > > Ideally allowing non-prefixed options should become a deprecated 'feature' > and eventually obsoleted, thus mandating that options be fully qualified. > Thoughts? > > Thanks, > Paul > > > > On Sat, May 11, 2013 at 12:14 AM, Christian Posta < > christian.po...@gmail.com > > wrote: > > > inline... > > > > > > On Fri, May 10, 2013 at 6:44 PM, Paul Gale <paul.n.g...@gmail.com> > wrote: > > > > > Hi Christian, > > > * > > > >What client are you using to connect with JMS?* > > > Java JMS, .NET and Ruby based STOMP clients. I'm only involved with the > > > Ruby clients. I have no idea how a JMS client is coded. > > > > > > Are all of these options reserved exclusively for non-STOMP clients? > > > http://activemq.apache.org/tcp-transport-reference.html > > > > No. those configure the TCP transport. They are not openwire or STOMP > > exclusive. They are also not client or broker specific. I can update the > > wiki, but they can more explicitly be set as "transport.soTimeout" or > > "transport.soLinger"... > > > > > > > > > > > Are all of these options available for both clients and brokers? > > > http://activemq.apache.org/configuring-wire-formats.html > > > > Yes, these configure the wire-level protocol options (openwire). They are > > specific to openwire. > > > > > > > > > > > Lastly, can these options be specified on a broker's non-STOMP > transport > > > connector URI (I presume they're illegal on a STOMP transport > connector)? > > > > > > transport.ioBufferSize=16384 > > > transport.maximumConnections=1000 > > > transport.soTimeout=10000 > > > transport.soWriteTimeout=10000 > > > transport.socketBufferSize=131072 > > > transport.threadName > > > > > Yes. But regarding your question previously asked... those options are > > reserved for client-side settings. > > > > > > > > During startup, it would be nice if ActiveMQ logged the options, parsed > > > from the transport connector's URI, that will be honored and warned > about > > > those that will be ignored. I read somewhere that incorrect options are > > > 'silently' ignored. Why silently? Log a warning for each ignored > option! > > > > > Thanks for your feedback. I will double check the debug-level settings > and > > add this if they don't already to that. > > > > > > > > > > Sorry for asking multiple questions in a single email: multiple answers > > are > > > much appreciated. > > > > > > > good questions :) no worries. > > > > > > > > Thanks, > > > Paul > > > > > > > > > > > > On Fri, May 10, 2013 at 7:10 PM, Christian Posta > > > <christian.po...@gmail.com>wrote: > > > > > > > s/JMS/the broker > > > > > > > > > > > > On Fri, May 10, 2013 at 4:09 PM, Christian Posta > > > > <christian.po...@gmail.com>wrote: > > > > > > > > > So you said transport connector... as in broker side... those > options > > > are > > > > > for client-side connections and apply to the > > ActiveMQConnectionFactory. > > > > > What client are you using to connect with JMS? > > > > > > > > > > > > > > > On Fri, May 10, 2013 at 7:49 AM, Paul Gale <paul.n.g...@gmail.com> > > > > wrote: > > > > > > > > > >> Hi, > > > > >> > > > > >> Can the following options also be used on a STOMP transport > > connector? > > > > >> > > > > >> watchAdvisoryTopics, alwaysSessionAsync and dispatchAsync > > > > >> > > > > >> Thanks, > > > > >> Paul > > > > >> > > > > > > > > > > > > > > > > > > > > -- > > > > > *Christian Posta* > > > > > http://www.christianposta.com/blog > > > > > twitter: @christianposta > > > > > > > > > > > > > > > > > > > > > -- > > > > *Christian Posta* > > > > http://www.christianposta.com/blog > > > > twitter: @christianposta > > > > > > > > > > > > > > > -- > > *Christian Posta* > > http://www.christianposta.com/blog > > twitter: @christianposta > > > -- *Christian Posta* http://www.christianposta.com/blog twitter: @christianposta
