[Kolab-devel] Yes, IMAP is good! (Re: Is IMAP still the right choice?)

[Bernhard Reiter]

Hi Thomas,

On Monday 21 November 2011 10:41:58 Thomas Koch wrote:
> I'll push regular updates to my bachelor thesis[1], if my Professor agrees.
> Right now I've pushed a list of advantages (short) and disadvantages (long)
> of using IMAP for a groupware like Kolab does.[2]

first I like that you are doing your thetis about Kolab and that you are able 
to share your results with us. Unfortunately, years of development with 
discussions about the design ideas are hard to summarize. If you get a chance 
you should talk to Martin. He is a busy person and the master mind behind the 
Kolab Concept.
 
> I've the impression that IMAP may have been a good, quick choice at the
> beginning of the project that should be revised now.

I am still convinced that IMAP is the right choice. As others, especially 
Georg, have pointed out, it is a specialised database for smart clients that 
enables to scale to hundereds or millions of users in one installation with IO 
bound, "dump" servers.  I agree with them.

For the same reasons I do not see the future in more helping functions on the 
server. Libraries for clients can be very useful, but the best optimisation 
will be within a smart client that uses the full potential of IMAP and its 
"disconnected" capabilities across the full vertical stack. The groupware 
challenge is less on the server side or standard formats, if you ask me, but 
about creating really smart clients that can deal with many sources and 
destinations for its data.

Of course, there will be clients that will run a lot of their code and state 
keeping data structure on a "server" in order to enable dump frontends like 
modern one-threaded apps, but those clients would still be clients towards
the storage server based on IMAP.

Best Regards,
Bernhard 

-- 
Send with http://userbase.kde.org/Kontact_Touch from an ideapad running MeeGo