[Kolab-devel] Re: steffen: server/kolabd/kolabd/templates DB_CONFIG.slapd.template, 1.2, 1.3 amavisd.conf.template, 1.6, 1.7 clamd.conf.template, 1.2, 1.3 cyrus.conf.template, 1.1.1.1, 1.2 fbview.conf.template, 1.1.1.1, 1.2 freebusy.conf.template, 1.1.1.1, 1.2 freshclam.conf.template, 1.2, 1.3 httpd.conf.template, 1.6, 1.7 httpd.local.template, 1.1, 1.2 imapd.conf.template, 1.5, 1.6 imapd.group.template, 1.1.1.1, 1.2 ldap.conf.template, 1.1.1.1, 1.2 ldapdistlist.cf.template, 1.1, 1.2 ldaptransport.cf.template, 1.1, 1.2 ldapvirtual.cf.template, 1.1, 1.2 main.cf.template, 1.18, 1.19 master.cf.template, 1.12, 1.13 php.ini.template, 1.4, 1.5 proftpd.conf.template, 1.1.1.1, 1.2 rc.conf.template, 1.2, 1.3 resmgr.conf.template, 1.7, 1.8 saslauthd.conf.template, 1.1.1.1, 1.2 session_vars.php.template, 1.4, 1.5 slapd.access.template, 1.1, 1.2 slapd.conf.template, 1.14, 1.15 slapd.replicas.template, 1.1.1.1, 1.2 smtpd.conf.template, 1.1.1.1, 1.2 transport.template, 1.1.1.1 , 1.2 virtual.temp late, 1.1.1.1, 1.2

[Stephan Buys]

Hi,

Humble apologies if the mail had a sarcastic tone. It was true excitement and 
not meant as criticism. I have the greatest respect for your work on the
server.

On Saturday 30 July 2005 02:27, Steffen Hansen wrote:
> We're happy for your contribution of course, but given the slight 
> sarcasm in your mail, I have to say it didn't really work until now.
> 
Makes sense, it was never really tested much. It was a feature we implemented
and then put on hold while Kolab2 stabilized.

> The few lines of code to parse the meta-data part of the templates are 
> fine of course, but getting things to work with the templates that we 
> handle in special ways (like imapd.group) did require more than that.
> 
I don't think we ever even considered all the special-case files like imapd.group.

> Anyway, I'd like to push the "update command" out to the templates also 
> if possible. I haven't experimented with it yet, but I fear that just 
> having
> 
> COMMAND=/kolab/bin/openpkg rc foobar restart
> 
> in the templates is not good enough because we can't control the order 
> of things. Also it'll be difficult to prevent multiple restarts of the 
> same service unless the COMMAND given in the templates are byte-by-byte 
> equal. Any good ideas?
> 
Pushing the "update command" out to the templates was also a wish-list item
for me. What about having a couple of variations?

ON_UPDATE_EXEC=/command/to/be/executed
  This will run a command whenever the templates differ.

UPDATE_TEST=/command/to/be/executed
  Implement the "test for change" as a script. Depending on the return-code
  of the script this will trigger "COMMAND"

ON_UPDATE_RESTART=imapd
  This will restart imapd using /kolab/etc/rc imapd restart
  We can do a lowercase comparison of the commands. This way spaces or byte-level
  differences wont make the system fail.


Kind regards,
Stephan




> regards