Bernd Fondermann wrote: > Noel J. Bergman <[EMAIL PROTECTED]> wrote: > > [Dynamic changes would be lost] until we move the configuration into > > JNDI (loadable from XML), at which point changes are easily managed, > > and amenable to clustering.
> but its a one-way-street nevertheless. all the nice comments from > config.xml get lost. There is no reason that they be lost. Many application servers provide perfectly nice user interfaces to dynamically managed configuration data. We could provide help, validation, standard docs/comments, and also preserve comments as an attribute of the configuration element. > would you as the administrator then re-start the server with the config > contained in the directory (not easily readable) or from file? *IF* we were to use a binary format, it could still be be perfectly readable with a myriad of tools, and we could provide an option to load from XML and replace or modify what is in the stored configuration. But please keep in mind that there is no reason why JNDI needs to write to a binary repository. It could write to XML as easily as anything else. See also: http://jakarta.apache.org/commons/configuration/. Although the current JNDI implementation there is read-only, also take a look at XMLConfiguration, which is read-write and uses XML files, therefore resolving the issue even without JNDI, albeit in a manner not quite as amenable to clustering. So I would move to use Jakarta Commons Configuration for JAMES, rather than the Avalon configuration, for both JAMES and the Mailet API (caveat: Danny suggests that we have our own, similar, configuration API in Mailet, and use Commons Configuration for the implementation). That was already on the Avalon roadmap before the projected imploded. --- Noel --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
