Soliciting Agenda for Weekly Server SIG Agenda (2015-11-10)

[Stephen Gallagher]

What do we want to discuss today, folks?
I think we should have a follow-up on the Cockpit criteria discussion. Additionally, I
think it would be good to select a subcommittee to work with the Cloud SIG to work on our
plans in that space. Lastly, I want to open a new topic: Should the Server SIG be
responsible for the "minimal" environment group in comps.xml? Technically, it is
available on the Server install media, so we should decide if that makes sense for us to
own in an official capacity. Many people already assume that we do, but as of right now,
minimal installs report themselves as non-productized installations.

[Major Hayden]

On 11/10/2015 05:34 AM, Stephen Gallagher wrote:
>  Lastly, I want to open a new topic: Should the Server SIG be
> responsible for the "minimal" environment group in comps.xml? Technically, it is
> available on the Server install media, so we should decide if that makes sense for us to
> own in an official capacity. Many people already assume that we do, but as of right now,
> minimal installs report themselves as non-productized installations. 
As someone who uses the minimal environment quite regularly, this is definitely something
I'm interested to discuss. ;)

--
Major Hayden

[Josh Boyer]

On Tue, Nov 10, 2015 at 8:14 AM, Major Hayden <major(a)mhtx.net&gt; wrote:
>  On 11/10/2015 05:34 AM, Stephen Gallagher wrote:
> > Lastly, I want to open a new topic: Should the Server SIG be responsible for the
> "minimal" environment group in comps.xml? Technically, it is available on the
> Server install media, so we should decide if that makes sense for us to own in an official
> capacity. Many people already assume that we do, but as of right now, minimal installs
> report themselves as non-productized installations.
>
>  As someone who uses the minimal environment quite regularly, this is definitely something
> I'm interested to discuss. ;) 
Isn't this something that we've repeatedly asked the Base WG to define
and maintain?

josh

[Stephen Gallagher]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 11/10/2015 08:38 AM, Josh Boyer wrote:
>  On Tue, Nov 10, 2015 at 8:14 AM, Major Hayden <major(a)mhtx.net&gt;
>  wrote:
> > On 11/10/2015 05:34 AM, Stephen Gallagher wrote:
> >> Lastly, I want to open a new topic: Should the Server SIG be
> >> responsible for the "minimal" environment group in comps.xml?
> >> Technically, it is available on the Server install media, so we
> >> should decide if that makes sense for us to own in an official
> >> capacity. Many people already assume that we do, but as of
> >> right now, minimal installs report themselves as
> >> non-productized installations.
> > 
> > As someone who uses the minimal environment quite regularly, this
> > is definitely something I'm interested to discuss. ;)
>  
>  Isn't this something that we've repeatedly asked the Base WG to
>  define and maintain? 
That is and always has been unclear. I had a conversation with some of
the Base WG during the systemd conference last week and they weren't
sure if that was their duty either.

My understanding of the division is that it's the Edition SIGs
responsibility to be defining actual deliverable artifacts (which in
my mind means any installable env group shipped on our media). The
Base WG's responsibility is to define the platform atop which the
artifacts can be built.

I don't (personally) think that the Base WG output should necessarily
be installable or even self-hosting; I think it should be a definition
of what constitutes the minimal set of functionality for which we are
willing to call something "Fedora". (Note, I intentionally choose the
term "functionality" and not "packages", because I think we also need
to consider the possibility that reorganization of the output of
packages may be necessary to produce a sensible view of this base
platform.

This may be a controversial statement about the Base WG. I think it
matches the definition provided at
https://fedoraproject.org/wiki/Base, however. (With the exception that
I don't believe Base should be self-hosting, because that necessarily
makes it a larger set than any of the installed Editions.)

> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v2
>
> iEYEARECAAYFAlZB+BsACgkQeiVVYja6o6PeEQCfWH/tuZ2nEFtJV+M8wzWF7i5W
> u0MAn0fT0bcOZGdn0ii+OmtvRE9czDF9
> =gmbC
> -----END PGP SIGNATURE-----

[Josh Boyer]

On Tue, Nov 10, 2015 at 8:58 AM, Stephen Gallagher <sgallagh(a)redhat.com&gt; wrote:
>  -----BEGIN PGP SIGNED MESSAGE-----
>  Hash: SHA1
>
>  On 11/10/2015 08:38 AM, Josh Boyer wrote:
> > On Tue, Nov 10, 2015 at 8:14 AM, Major Hayden <major(a)mhtx.net&gt;
> > wrote:
> >> On 11/10/2015 05:34 AM, Stephen Gallagher wrote:
> >>> Lastly, I want to open a new topic: Should the Server SIG be
> >>> responsible for the "minimal" environment group in comps.xml?
> >>> Technically, it is available on the Server install media, so we
> >>> should decide if that makes sense for us to own in an official
> >>> capacity. Many people already assume that we do, but as of
> >>> right now, minimal installs report themselves as
> >>> non-productized installations.
> >>
> >> As someone who uses the minimal environment quite regularly, this
> >> is definitely something I'm interested to discuss. ;)
> >
> > Isn't this something that we've repeatedly asked the Base WG to
> > define and maintain?
>
>  That is and always has been unclear. I had a conversation with some of
>  the Base WG during the systemd conference last week and they weren't
>  sure if that was their duty either.
>
>  My understanding of the division is that it's the Edition SIGs
>  responsibility to be defining actual deliverable artifacts (which in 
Yes, for _their_ Edition.  Not as a common denominator.

>  my mind means any installable env group shipped on our media). The
>  Base WG's responsibility is to define the platform atop which the
>  artifacts can be built. 
Where is that definition?  Or more succinctly, if that definition is
whatever is on the wiki page, what purpose is the Base WG serving
today as it seems to be "done" or otherwise stalled.

>  I don't (personally) think that the Base WG output should
> necessarily
>  be installable or even self-hosting; I think it should be a definition
>  of what constitutes the minimal set of functionality for which we are
>  willing to call something "Fedora". (Note, I intentionally choose the
>  term "functionality" and not "packages", because I think we also
> need
>  to consider the possibility that reorganization of the output of
>  packages may be necessary to produce a sensible view of this base
>  platform. 
I'm not comfortable with the minimal functional install being the
responsibility of any of the editions, particularly not Server or
Workstation.  Cloud is arguably the most minimal of the editions.
However, when you start talking about _function_, what is minimal is
completely unclear from one Edition to another.

That is not to say that Server hasn't done a good job as the de facto
maintainer of this.  I would simply prefer to non-Edition entity to be
responsible for the definition and curation of minimal function.
Perhaps Env & Stacks if not Base.

josh

[Stephen Gallagher]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 11/10/2015 09:14 AM, Josh Boyer wrote:
>  On Tue, Nov 10, 2015 at 8:58 AM, Stephen Gallagher
>  <sgallagh(a)redhat.com&gt; wrote:
> > -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
> > 
> > On 11/10/2015 08:38 AM, Josh Boyer wrote:
> >> On Tue, Nov 10, 2015 at 8:14 AM, Major Hayden <major(a)mhtx.net&gt; 
> >> wrote:
> >>> On 11/10/2015 05:34 AM, Stephen Gallagher wrote:
> >>>> Lastly, I want to open a new topic: Should the Server SIG
> >>>> be responsible for the "minimal" environment group in
> >>>> comps.xml? Technically, it is available on the Server
> >>>> install media, so we should decide if that makes sense for
> >>>> us to own in an official capacity. Many people already
> >>>> assume that we do, but as of right now, minimal installs
> >>>> report themselves as non-productized installations.
> >>> 
> >>> As someone who uses the minimal environment quite regularly,
> >>> this is definitely something I'm interested to discuss. ;)
> >> 
> >> Isn't this something that we've repeatedly asked the Base WG
> >> to define and maintain?
> > 
> > That is and always has been unclear. I had a conversation with
> > some of the Base WG during the systemd conference last week and
> > they weren't sure if that was their duty either.
> > 
> > My understanding of the division is that it's the Edition SIGs 
> > responsibility to be defining actual deliverable artifacts (which
> > in
>  
>  Yes, for _their_ Edition.  Not as a common denominator.
>  
> > my mind means any installable env group shipped on our media).
> > The Base WG's responsibility is to define the platform atop which
> > the artifacts can be built.
>  
>  Where is that definition?  Or more succinctly, if that definition
>  is whatever is on the wiki page, what purpose is the Base WG
>  serving today as it seems to be "done" or otherwise stalled.
>  
> > I don't (personally) think that the Base WG output should
> > necessarily be installable or even self-hosting; I think it
> > should be a definition of what constitutes the minimal set of
> > functionality for which we are willing to call something
> > "Fedora". (Note, I intentionally choose the term "functionality"
> > and not "packages", because I think we also need to consider the
> > possibility that reorganization of the output of packages may be
> > necessary to produce a sensible view of this base platform.
>  
>  I'm not comfortable with the minimal functional install being the 
>  responsibility of any of the editions, particularly not Server or 
>  Workstation.  Cloud is arguably the most minimal of the editions. 
>  However, when you start talking about _function_, what is minimal
>  is completely unclear from one Edition to another.
>  
>  That is not to say that Server hasn't done a good job as the de
>  facto maintainer of this.  I would simply prefer to non-Edition
>  entity to be responsible for the definition and curation of minimal
>  function. Perhaps Env & Stacks if not Base.
>   
So, I think that at the very least, Server should be responsible for
anything selectable from the Anaconda installer on the Fedora Server
install media. The problem here is that Anaconda will display any
entry from comps that is fully satisfied by the packages available on
the media.

So if we're going to offer "minimal" as an installable option, I think
we need to own it. Otherwise, I think we need to hack up a mechanism
in comps that makes it so that the Server isn't a proper super-set of
"minimal" so that it doesn't appear on the list.

> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v2
>
> iEYEARECAAYFAlZB/GEACgkQeiVVYja6o6NmIACeMOjaokVdYWnRXr4bFbGrhUam
> SWkAoKotvK3QTx4RdtG4KfUEDxv1JK7d
> =OnyA
> -----END PGP SIGNATURE-----