Bringing fdo.org to the next level

[Waldo Bastian]

On Friday 15 April 2005 12:17, Philippe Fremy wrote:
> If your proposition relates to the lengthy discussion about d-vfs or
> dconf, my very personal opinion is that you should first start to write
> code, and then show it and discuss it (like Linus said, "show me the
> code!"). 

I think that is well meaning advice, and having working code at some point is 
certainly the best way to get anywhere, but there is a very big risk when you 
start with code in relation to something like d-conf and that is that people 
will look at your code and assume that whatever design decision you have made 
in your code is final and then use that as a reason to dismiss the whole idea 
out of hand. I think the current process actually works fairly well in 
working out how this ideal new VFS or configuration system should look like. 

Even though most open source software starts with some rough code, the 
circumstances here on FreeDesktop.org are slightly different. For starters 
there is already enough working code that we can look at for reference. 
Second, even though the waterfall method isn't traditionally used very much 
in open source projects, it's isn't such a bad model to start with. Once you 
have done the first iteration you will have some sort of working code and 
then you can move to faster iterations that are more typical for open source.

The first code doesn't need to match all the requirements, but it really helps 
if you know where you are going, with a lot of software the requirements 
aren't at all clear and evolve over time but in this case there is a pretty 
good understanding among the various stakeholders what they expect out of 
this (And it better meet their expectations otherwise they are unlikely to 
use it) The major challenge is to get input from all stakeholders and make 
sure that the overall requirements correctly reflect all of the individual 
requirements. 

When that has happened you will hopefully be able to sell the idea to the 
various projects and hopefully get a balanced involvement in the 
implementation phase from the various stakeholders. 

I think the latter is something that more or less failed with DBUS at the KDE 
side. Let's see if we can do better this time.

Cheers,
Waldo
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://lists.freedesktop.org/archives/xdg/attachments/20050415/b782460e/attachment.pgp