Re: [gentoo-server] QA or an unchanging portage tree?

[Rodney Amato]

I have never really looked into how portage is made up to deeply but I 
would imagine that doing
number 2 would have the most benefit and if you are using cvs as I think 
you are then you might be
able to just have a different branch in the tree that has like a 6 month 
life span. By default you would get
 the bleeding edge stuff so that it gets tested more but for those who 
needed a slower moving target
they could just change the module they are rsyncing too from 
gentoo-portage to gentoo-stable-portage
or something like that.

Provided the mirrors don't mind the extra space usage number 2 is going 
to be the way to go because
it is going to be easier to implement IMO and have the most benefits. 
The part I am not sure about is
at what point would something qualify to make it into stable ?

Kurt Lieber wrote:

>All --
>
>I'd like to poll the group to get your input on a question that has come up
>recently.
>
>There are a number of areas where Gentoo Linux could stand improvement --
>we all know this.  Two examples being discussed now are a) improved QA for
>the portage tree and b) the fact that the portage tree is very fluid and
>dynamic, which makes it more difficult to use in enterprise environments.
>
>If you were given the choice between:
>
>1) A more robust QA process for the main portage tree or
>2) A seperate 'server' portage tree that offered:
>   * only updated quarterly
>   * security and major bug-fixes off-cycle, but no other changes to the
>     tree
>   * guaranteed minimum life of all ebuilds in the tree
>
>Which would you find more valuable and why?
>
>--kurt
>
>

-- 
Rodney Amato
Systems Administrator
http://www.squiz.net
>> 92 Jarrett St Leichhardt, Sydney, NSW 2040 ...>
T: + 61 2 9568 6866
F: + 61 2 9568 6733
.....>> Open Source  - Own it  -  Squiz.net ...../>