Michael's FUD (Was: Fragen zu Synaptic)

[Ronny Buchmann]

On Sunday 13 June 2004 16:31, Michael Schwendt wrote:
> In one message board (don't remember its name of the top of my head),
> someone described the difference between the release-cycle of fedora.us
> and 3rd party repositories like this:
>
>   fedora.us : new package -> fix, fix, fix, fix -> release
>   3rd party : new package -> release -> fix -> release -> fix -> release ->
> ...
>
> Which seems to hit the nail on the head. Though, nowhere do I claim that
> this is true for every package.
This is actually a really good summary of the working style, but it should be 
more detailed:

fedora.us: new package -> fix (now 90%perfect), fix (95%perfect), 
fix(97%perfect), fix(98%perfect) -> release

3rd party: new package (80%perfect)-> release -> fix (now 90%) -> release

may I quote you: "nowhere do I claim that this is true for every package." 

Often 3rd party packages start with more than 80%, and often fedora.us 
packages may end with 100% (if one doesn't negate 100% being possible at 
all).

But I think most users are well satisfied with 80-90%. And everyone knows that 
the last percents are the most expensive.

I don't mind if some people like to put that much effort into packaging QA (as 
in fedora.us, which is from my impression higher than inside RH, for example 
complete BuildRequires were not very common). But please don't bash others 
for not doing the same.

Fedora.us could be much better if it adopted parts of the 3rd party style. 
No/low QA for testing/unstable repo, and a user voting system to move 
packages to stable. (I'm suggesting this again, repetition seams to commit 
things to memory and sometimes even minds change)

-- 
http://LinuxWiki.org/RonnyBuchmann