[Desktop_architects] Portland: The Linux DesktopIntegrationInterface

[Martin Konold]

Am Montag, 5. Dezember 2005 19:42 schrieb Mike Shaver:

Hi,

> Version skew and dependencies only go away over time if you have a  
> "stable" (in the biological sense) set of capabilities.  I don't  
> think we aspire to that.

Yes, version skew _never_ goes away. There are imho only two scenarios:

1st: The platform is developed. In this case the version skew problem is 
preserved because as soon as some old version skew issue is solved some new 
version skew problem pops up.

2nd: The platform is stagnant which basically means it is dead. In this case 
the respective stakeholders most probably have no interest in solving the 
final version skew issues. So it is an expected outcome that the version skew 
is preserved just in the moment when the platfrom is considered dead.

IMHO a technical answer to this challenge is to go away from static interfaces 
and use more dynamic late "binding" interfaces and protocols. The later can 
be leveraged to provide more easy and flexible upgrade paths without breaking 
existing installations or creating incompatible installations. 

Just look how well the internet protocols did develop during the last two 
decades. IMHO this is largly due to the fact that it is based on protocols 
not method calling conventions......

The later is already a problem on the Windows platform. E.g. I know of cases 
where different applications require different versions of dlls or 
environments which are not offered by MS for parallel installations in 
general.

I want to avoid this problem for the Linux Desktop without parallel 
installs(*) of different versions of the same service provider.

Regards,
-- martin
(*) Parallel installs are only band-aid like workarounds instead of a solution 
of the problem. It can easily happen that you need many parallel installs 
with many different ISV offerings available for the Linux Desktop 
ecosystem....

-- 
http://www.erfrakon.com/
Erlewein, Frank, Konold & Partner - Beratende Ingenieure und Physiker