Hmmm.

First mistake: Upstream only offers building blocks but no integrated system – that’s left to downstream.

Second mistake: Downstream dicks around with code from upstream.

Third mistake: ?

“€20 per month is quite a lot.”

For you, mayhaps. But for a corporation paying €thousands per employee and month, no, it isn’t. You might as well start worrying about the toilet-paper budget.

ch
“And why in the world would I use the Enterprise version for 25 users? So the price for Exchange is $708, and your calculations go out the window. Even if you install two servers, that’s just $1416.”

Out box that only support 5 user/devices add on 20 68×20 that is another $1360 in cal to get up to 25 users. $2781. Basically Zimba Standard for 6 years just with support or 3 years with a Perpetual. That is as long as you only stop at 2 servers.

ch no matter how you look at it that instance pricing is a problem.

Now I do the maths on standard you could say almost 1 for 1 but appears to be in exchange standard avantage.

There is a problem here Exchange standard cannot be put head to head with Zimbra standard or even Zimbra basic. All Zimbra support integration with asterisk. So this “Voicemail with Unified Messaging” that you only get in enterprise exchange.

Also per device passwords are default on all Zimbra for activesync that also only comes in enterprise exchange.

Accepting the fact that Exchange standard and Zimbra standard don’t match brings a problem on compare how much each of those is worth.

For the best feature match against Exchange Standard is Sogo. The problem its it costing system is different.

http://www.sogo.nu/english/support/commercial_support.html

$5000 dollar per year no matter how many servers you have. No cals or mailbox counting. As many as you hardware can handle. I know there are cheaper support options there for single server I would not be touching those because I am going to deploy pairs at min. So any single server failure does not cause any disruption.

The free version of Sogo is fully featured. Sogo there is no problem if you don’t pay. Don’t pay don’t call support. In fact having 24/7 support call with Microsoft costs extra. Mostly works out worse than paying the $5000 or $10000.

Note the €6000 for 25 users. These no user counting licenses that are out there can be quite cheap to large companies.

Ch Zimba is about the most expensive in the FOSS class you should go for. Open-xchange is also up there for being expensive it also has instance pricing why I never touch it.

The most expensive for the features you get in group-ware is Exchange bar none. If you start seeing that something is more expensive than exchange standard normally the problem is it has extra features that exchange standard does not have.

Ch something has instance pricing start worrying basically. This now means when you can cluster something because you have the free hardware resources todo so you don’t because you cannot get approval for the extra license.

Ch old saying “No such thing as too many backups. There is only such thing as one too few.” Wrong licensing forces you into one too few way too often.

This is the problem in most Windows networks you walk into how many servers 1. What happens if that server fails business is in big trouble. Most of this is caused by the instance pricing of the software not the cost of the hardware.

In one way I am glad to see MS kill off the small business server that prick would not allow multi ads in network.

At 100 users at MS costings you are normally cheaper to go out and buy the support contracts for the instance free, count free FOSS stuff. Less paperwork less headaches.

This is when you are trying to go cheap.
http://kolabsys.com/pricing
70 dollars for a 50 user server.
200 dollars for 50 active sync.
800 dollars for 50 client software kolab client is their outlook replacement.
450 dollars if you choose to go Thunderbird.

So if you skip there client. 70+200=270 per server cost. 1 exchange standard server with 5 cals you can buy a server that handles 10 times as much. At 2 servers you can afford to buy the PIM client(ie the 800 dollars) yes you do have room for a few outlook connectors.

Of course you can choose not to run the certified versions and use kolab completely for free with no user limits.

When you know the price of kolab. Zimbra is expensive.

kolab is that cheap you cannot buy less than 50 users. They will not sell you a 5 or 25 user license.

Every 4 cals on an exchange standard is the cost per server for a extra year on kolab. The instance of exchange standard price buys you 2 years straight up. By the time you get to 25 users dual server kolab is cheaper. When you get to 50 its way cheaper. At 50 even spending on all the optional extra 37 dollars per named user for a single server. To get to the 68 dollar cal price using kolab you have to be running 3 servers. 110 of enterprise is 6 servers of kolab. That is to match the cal price. Remember you have got your client side software for that. By cal windows exchange enterprise instance is only worth 2750 yet MS is charging $4051 and standard is only worth $340. So you are handing over a decent amount in instance charge.

Using other providers you don’t pay these instance charges.

Brillo
“Unit tests are supposed to be done way, way before intergating the application to the system. If you find yourself running unit tests while deploying the app, you are do it wrong.”

History of failures tell you repeatily if there is a produced binary and an extra step has to be done to perform unit tests human will screw up and skip doing the unit tests.

ch
“Please explain how the best unit test in the world could have found out that particular problem?”
I said a lesson learnt from the openssl disaster.

We you study why something failed you check out every path. Now if openssl had a unit test that could have detected it since the unit tests had never been run would have changed nothing.

So to correct the problem that happens with openssl was alter how unit tests run and implement a unit test. Because 3 mistakes made the openssl disaster happen.

Openssl disaster is one of the most recent but going backwards there were before that a few glibc disasters caused by unit cases not being run on build and a few gcc disasters caused by the same thing. Openssl is just the most recent example of a software builder skipping over the unit testing.

Even MS products have come out and failed there own provided testing. Its not just a FOSS thing. There is only 1 way todo unit testing basically correctly in the build design of the application to make sure it happens most of the time. On by default.

Yes you are right is meant to happen. History of human errors tells you application build process directly effects how dependable that it will be done before you get the application. From a safe design build process OpenOffice is not.

Please explain how the best unit test in the world could have found out that particular problem? The code did exactly what the programmer wanted it to, and it did produce random numbers as intended – so it would pass all tests. It would take more than a unit test to find out that the random numbers are not random enough.

So, to quote Brillo:

There is no end to your made-up nonsense, is it?

Unit tests are supposed to be done way, way before intergating the application to the system. If you find yourself running unit tests while deploying the app, you are do it wrong.

There is no end to your made-up nonsense, is it?

Who says it’s either – or? They do pay christmas bonusses

“The only thing worse than giving M$ money would be throwing it to the wind”

Not true for businesses, but you don’t want to understand that, of course.

“(waste and littering)”

Don’t worry, that money won’t litter the landscape for long

Well, I suppose that’s a good thing if your company has nothing better to do with €20… Let’s see… Free Lunch every Wednesday, breakfast every day, Christmas Bonus, … The only thing worse than giving M$ money would be throwing it to the wind (waste and littering).

And why in the world would I use the Enterprise version for 25 users? So the price for Exchange is $708, and your calculations go out the window. Even if you install two servers, that’s just $1416.

And if my company is big enough to warrant the Enterprise version, my procurement staff will have a nice ol’ chat with MS’ salesmen. In the really large company I worked for the combined price for all the MS products used was on the order of €20 per employee and month (Chris Weig, can you help me out with the exact number?) – roughly what some companies pay to supply their employees with soft drinks.

For a large number of configurations Zimbra open source offering has lots of features for $0. e.g. most of the features are identical with the pricey versions (7 pages of an 11-page list of features). Just at a glance, the differences seem to be multiple domains and mobile syncing. Those don’t strike me as must-haves for any small or medium-sized organizations. It might matter for organizations with large numbers of mobile staff or global corporations enduring mergers… rather small niches.

Brillo its different now. You brought up the PSP as an example. Really go read what Sony will and will not license for the PSP.

http://www.microsoft.com/exchange/en-us/pricing-exchange-server-email.aspx
Ok since you asked for you head to be ripped off.
Brillo $4051 per instance price of exchange enterprise.

http://www.zimbra.com/products/pricing.html
$2,231 / 25 pack 3 year subscription.

So you want to run exchange enterprise with a live backup 25 cal you are straight up for 2 instances that is a nice $8102 dollars. Now if I want to run 4 instances $16202. That instance price on Exchange Enterprise becomes a real expensive bastard really quickly. That instance cost stacks up big time.

6 years zimbra professional only puts you back $4461 for as many servers you want clustered for reduncey. Remember you only have 5 years with exchange by life cycle before having to pay extended support. So $4461/6×5=3717.5 is the correct price for Zimbra 1 server about the same 2 servers you are saving 4 you are ahead by miles.

So almost half for my case were I install with backup servers is true. The cal price on exchange enterprise is $110 a cal. Zimbre 6 years $60 dollars. Lets say I throw in a Perpetual for good measure $63 then it comes upto slightly more expensive at $120 but you have 1 extra year of support. That $120 is high because the support alone is less than the multi year license. Then cost per named user price comes about the same.

But you have to remember you never need to rebuy the Perpetual part once you have a Perpetual of X number of mailboxs that is yours forever across all future versions. Upgrades are linked to your last Advantage/Premium Support contract.

First cycle Zimbra can work out about the same as exchange if you only run 1 server and buying the Perpetual as well as support. Next cycle its about 50 percent. Now if you want some true redunacny straight off the bat Zimbra is cheaper by large margin particularly when you want a lot of redunancy.

Offical mainstream support for Exchange is 5 years. Past that you have to repay anyhow.

I was nice about dmake but since you want to fight you asked for it Brillo. Its clearly documented on the openoffice site about the problem.
http://www.openoffice.org/tools/dmake/
You don’t know dmake its unsupport trash. There should be a tool to take dmake to something that is maintained. About the only thing still using dmake is OpenOffice and Libreoffice. Libreoffice is in the process of getting rid of it.

Please note the dmake used by OpenOffice and LibreOffice is not Distributed Make(dmake) that provide in Solaris (Sun) Studio. The OpenOffice dmake is “Workshop dmake” in fact Workshop dmake will not do a Distributed build. Wrong dmake Brillo. Yes both makefiles have incompadible syntax. One is supported one is not. If it was using Distributed Make it would still be somewhere near sane.

When I said strange make system called dmake I did really mean that. You say dmake and the first thing people think of is Distributed Make. Not the strange one you have to use.

So basically OpenOffice is building with a program that is lead developer is gone and that is barely being maintained.

Brillo
–“*make” has essentially nothing to do with unit tests but will run the code for you as long as you have them specified in your Makefile.–

The unit test thing is an extra. It has found if you place unit tests as something option for a person todo they never run them. This traces to the openssl disaster of debian. Lesson of openssl disaster says making unit tests automatic as part of the build process that person building has to pass instruction to turn them off not on.

Brillo
“No one wants to develop and maintain an entire productivity suite in-house. That’s why they go out and buy that stuff in the first place. dmake? GNU make? WHO. CARES.”

Correct no one does what to maintain an entire productivity suite alone this is true. If you are apply a fix to something you don’t want want to be using a no longer support make system that you have to fix the make system just so you can build the program with the fix you need. Dmake(Workshop) should be long gone from Openoffice it has not been maintained properly for years. It should have been replaced by a maintained make system there are enough of those. Libreoffice has chosson to replace dmake(Workshop) with gnumake.

Brillo
“Give people something that works or don’t.”

The auto unit test one applies to this one exactly.

So when program builds and it says complete it works. Not have some extra step to find out that the program works. Extra step to run unit tests makes the possiblity of deploying a program that could have been detected as broken by the unit tests so preventing people like you getting a broken program.

Yes the unit test one is critical to quality production.

Brillo something built with a non maintained make system is sign of major problems particularly if there is no plan to properly maintain it or replace it.

Brillo you just proved you did not have a clue how bad OpenOffice is internally.