Re: Debian Pure Blends on www.debian.org
Jonas Smedegaard <dr@jones.dk> writes:
> Quoting Ole Streicher (2016-01-25 14:49:25)
>> Jonas Smedegaard <dr@jones.dk> writes:
>>> One detail: That new page describes metapackage as a minimum
>>> requirement for a blend - I disagree that to be a requirement¹, so am
>>> curious where that came from.
>>
>> In my reading this page does not describe a requirement, but a status.
>
> Here is the parts I am talking about:
>
>> "Released" may have different meanings for different blends. At a
>> minimum, it means that the blend has metapackages that were released
>> in a stable release of Debian [...].
>
> What does "at a minimum" mean in the context of "as we can see below
> they all currently happen to be expressed as metapackages"?
I have read in in the way you explain here with the context.
>> And currently every released Pure Blend *has* metapackages, right?
>
> Uhm, no. DebianParl exists and is deployed from what Is now in Debian
> stable (it was deployed _before_ stable became stable, and the actual
> deployment uses what is now oldstable - which Boxer can produce by use
> of purely stable packages - none of which are metapackages).
OK, didn't know that. However, youself state that it is not "Pure".
>> "The released Pure Blends have metapackages in the <a
>> href="https://www.debian.org/releases/stable/>stable</a> release of
>> Debian. Some Pure Blends also provide installation media or form the
>> basis of a derivative distribution. See the individual Blends pages
>> for more information."
>
> What do we gain from first sentence above (apart from confusing some
> readers - including myself - wrongly reding it as a definition rather
> than an observation - which then may or may not be true in the
> future)?
>
> How about dropping first sentence from above?
When it is not true, it should be changed. It is written as the landing
page for newcomers, and they (I would say) care about the status, not
the definition. The status may change, and the web page should change as
well then.
> ...and the second sentence I disagree with: A blend is not pure when
> providing installation media which is not the Debian installation
> routine itself.
So if I provide an (installable) live-Image of Debian-Astro, I would
"un-pure" the Blend?
>> A naming question: are there any "non-pure" blends yet?
>
> Yes: DebianParl is currently non-pure, in that it tweaks a few
> configuration files (see bug#311188 for that kind of problem).
I see a contradiction between the goal of a Debian Pure Blend -- namely
to provide a system "that is configured to support a particular target
group out-of-the-box" (Wiki definition of a Pure Blend) and not tweaking
the configuration.
But if DebianParl is unfit for being a released Pure Blend, it can't be
taken as example for the first section of the web page.
Would it be a solution to add a section for Non-Pure Blends with
DebianParl and DebianEdu? (and FreedomBox?) This may also have the room
to explain why they are not Pure.
> IMO we should use only the terms "Debian Pure Blend" and "Debian Blend",
> and only the way they are defined, not try elaborate/simplify their
> purpose.
Ack.
> If too hard to grasp the two terms, then I agree we should drop the
> non-pure term (*not* drop the pure form, as Andreas has proposed in the
> past).
If we drop the non-pure term, then DebianParl and DebianEdu are not
Blends anymore, right? Would sound a bit silly to me...
Best regards
Ole
Reply to: