I'm not suggesting an actual change of ports while running, just changing the configuration files. There would be an ofbiz reset involved. Just when it comes back up, if it's not changed concurrently, it won't be accessible.
----- Original Message ---- From: David E Jones <[EMAIL PROTECTED]> To: dev@ofbiz.apache.org Sent: Saturday, December 15, 2007 12:38:42 AM Subject: Re: OfBiz System Configuration Wizard Wow, I didn't even realize we were considering something to change ports on the fly. Has anyone even done a proof of concept to see if the various infrastructure pieces we're using even support that? I guess in theory they should, but you're getting into a LOT more than just reading and writing files or something... you may have to unload and reload different objects and everything that depends on them.... Beyond a proof of concept that would also have to be tested a lot because those tend to be error prone sorts of things, especially when the infrastructure was not originally written with that in mind. -David On Dec 14, 2007, at 10:58 PM, Chris Howe wrote: > To use the http port setting again. If you're using the UI to > change the port and you only change one of the files, you likely > will have to go back and change the file by hand after a restart > because the UI won't be accessible. Whereas if it were changed in a > transactional manner it either fails and you're presented with the > same UI or it passes and the new settings take effect properly. > > Also in the case of conflicting changes. Beanshell, email, http, > etc all need to be running on different ports. If you happen to > have them running on a port that Ofbiz already has in use, you > likely won't be able to get back to a UI to correct the mistake. In > a transactional manner, you're able to run a service to verify such > things before committing it back to the file system. > > ----- Original Message ---- > From: David E Jones <[EMAIL PROTECTED]> > To: dev@ofbiz.apache.org > Sent: Friday, December 14, 2007 11:35:33 PM > Subject: Re: OfBiz System Configuration Wizard > > > > The transactional nature sounds wonderful, but what problem does it > actually solve? > > -David > > > On Dec 14, 2007, at 10:32 PM, Chris Howe wrote: > >> d) Load the config files into an XML database (Apache Xindice) >> manipulate with a UI/wizzard to your heart's content, verify the >> structure against an xsd, flush it to the original filename. The >> benefit of using an xml database as opposed to just reading/writing >> the original file is that you're able to make the changes in a >> transaction manner. For instance changing the http port, you can >> change url.properties and ofbiz-containers simultaneously. >> >> ----- Original Message ---- >> From: Jacopo Cappellato <[EMAIL PROTECTED]> >> To: dev@ofbiz.apache.org >> Sent: Friday, December 14, 2007 11:08:01 PM >> Subject: Re: OfBiz System Configuration Wizard >> >> >> I think that system settings should stay in config files, not in the >> database; if the goal is to simplify the configuration steps > described >> in the production setup guide, then there are probably different ways >> of >> addressing this: >> >> a) deliver a separate set of config files already configured for a >> "standard" production (cache enabled, verbose logs disabled etc...); > >> we >> >> may also consider to deliver these settings in the release branch, > and >> maintain the dev settings in the trunk >> >> b) implement an ant-based wizard (to be run during the installation > of >> OFBiz) that prompts the user for some common settings (http port, >> https >> >> port, mail server address, db used, db user/password, db url etc...) >> and >> then modify the OFBiz's files (or, we could prapare *one* simple file >> where the user can enter all these values, then run the ant script >> that >> >> places then in all the relevant OFBiz files) >> >> c) clean up the existing config files; for example, the >> entityengine,xml >> contains the settings for a lot of different databases; we could keep >> the settings for just one of them and move the others into a separate >> file, or create one file per database etc... >> >> Of course, for more complex (real World) setups, you'll have to > follow >> the steps of the production setup guide... but for simpler ones it >> could >> work. >> >> just my 2 cents >> >> Jacopo >> >> >> > > > >
