----- Forwarded Message -----
From: Martin SolĨiansky <[email protected]>
To: [email protected] 
Sent: Friday, February 15, 2013 2:56 AM
Subject: zimbra
 

hey,
I am trying to get a complete port built.

I've seen the incredible effort that you put in and I believe it to be VERY 
CLOSE to a complete port build.

I have been in contact with the FreeBSD ports people and some have expressed an 
interest in getting this to be a complete and sanctioned port.

Please tell me that you're interested in finishing this to completion.
once again, zimbra forums notification confuse me. i got private message 
notification on feb 15th, 6 days later :(.
i can tell you that zimbra on freebsd is the same zimbra you can run on linux 
nowadays. university mailserver has been happy for some time now :). however, 
with a complete port build there are some major issues:
1) '/opt/zimbra' is hardcoded. seriously, hardcoded in lots and lots of parts 
of code. you can do a patch that will change this but it's very very bad idea 
in the long run. the maintenance would kill you. putting stuff outside /usr 
does not comply with freebsd standards. you can circumvent via mount_null but i 
doubt that freebsd people will approve it.
2) some libraries/versions, perl modules are simply NOT available in the ports 
collection and if they are there you just cannot reuse them inside the base 
system since zimbra hardcodes stuff into them. there is a reason why zimbra 
team uses specific versions of the third party software and i somehow learnt 
the hard way :-). not to mention linking.
3) port upgrade would be a simple nightmare. you have to use the provided 
zimbra-upgrade script (or spent ages to do it your own way) and upgrade simply 
deletes all the binaries.. and replace them.

that being said, i am well capable of creating an initial port (hell, i started 
that long time ago and then just gave up) but i would need constant help. i 
don't have time to maintain it. and maybe i even lack some skill to do that. 
and i don't think that it can be done "properly".
some time ago i offered zimbra team i would rework their build process so that 
it becomes more.. transparent. well, i don't have the time now but prior to 
creating the port it would help a great deal if the zimbra team could just 
incorporate slight changes (include placeholders for the patch in the code, 
create ONE CONFIGURATION FILE FOR EVERYTHING (yes, caps)). that way you could 
do only slight alterations. simply change "master dir" and you are done (tbh, 
they use $ZIMBRA_HOME variable most of time.. and then.. they just switch to 
/opt/zimbra :))

so there you got your very first step. for several years the patch just "added" 
support for freebsd compilation while changing nothing for other platforms. i 
gave up in the last release and just hardcoded it 'cos once again i lack time. 
contact freebsd ports people, tell them to ask zimbra people to offer some help 
(they very friendly pre-vmware - dont know now).
the second step is finding someone who is interested in maintaining the port. i 
can do the patching for new versions but i simply won't handle the agenda 
(compatibility with other freebsd versions, compilation, alteration bla bla 
otherworsending in -ion).

i will try to prepare the newest zimbra version (8.0.2) patch and we can try 
the to make it happen :-). thing is.. i have no fucking idea why zimbra can't 
just do a proper supported commercial edition. it was proven that it runs on 
freebsd and it's somehow a very silly commercial model to exclude shitload of 
freebsd servers from their portfolio.

let's cure the disease, not the symptoms.

you can reach me on skype via solko_the_demigod handle. best of luck to you,

s.

PS (sorry about the html email but i just don't think that in these days anyone 
uses device without html support - and that comes from someone using fbsd for 
past 13 years)
_______________________________________________
[email protected] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "[email protected]"

Reply via email to