Stopping the mandatory buildroot insanity

[Axel Thimm]

So we voted on a new *mandatory* buildroot. By now people saw that the
previous *mandatory* buildroot entered the guidelines and started
blocking new packages requiring the old *mandatory* buildroot.

I don't know what fesco did last week on ratifying or not the new
buildroot, and either way people will think differently on any single
buildroot. Perhaps buildroots are the most unimportant piece of s**t
with the most polarized parties.

Now who's idea was it to have a *MANDATORY* buildroot at all?</rhetoric>

Most of the FPC are fed up and have often stated that the buildroot
guidelines should be simple "If it works, have it".

Plain and simple:

  Request for voting on dropping the *mandatory* from the guidelines
  and explicitely cast it into a *suggestion*

+1

The first other five positive voters get a free beer when I meet
them (we never said that committee members could not be bribed, or do
we need a guideline for that? ;)
-- 
Axel.Thimm at ATrpms.net

[Jason L Tibbitts III]

> >>>> "AT" == Axel Thimm
> <Axel.Thimm(a)ATrpms.net&gt; writes: 
AT> Request for voting on dropping the *mandatory* from the guidelines
AT> and explicitely cast it into a *suggestion*

-1

 - J<

[Toshio Kuratomi]

On Tue, 2007-02-20 at 09:20 -0600, Jason L Tibbitts III wrote:
>  >>>>> "AT" == Axel Thimm
> <Axel.Thimm(a)ATrpms.net&gt; writes:
>  
>  AT> Request for voting on dropping the *mandatory* from the guidelines
>  AT> and explicitely cast it into a *suggestion*
>  
>  -1 
-1
This puts us back where we started.  It was a suggestion until recently.
The fact that it wasn't mandatory just confused reviewers and made
people debate the issue over and over inside of bugzilla.

Let's move forward:
1) Make a clear rule on a buildroot value that fixes the technical
issues.
2) Get changes made to rpm so that F7/8 and RHEL6 don't have to worry
about this.

-Toshio

[Axel Thimm]

On Tue, Feb 20, 2007 at 08:57:48AM -0800, Toshio Kuratomi wrote:
>  On Tue, 2007-02-20 at 09:20 -0600, Jason L Tibbitts III wrote:
>  > >>>>> "AT" == Axel Thimm <Axel.Thimm(a)ATrpms.net&gt;
> writes:
>  > 
>  > AT> Request for voting on dropping the *mandatory* from the guidelines
>  > AT> and explicitely cast it into a *suggestion*
>  > 
>  > -1
>  
>  -1
>  This puts us back where we started. It was a suggestion until recently.
>  The fact that it wasn't mandatory just confused reviewers and made
>  people debate the issue over and over inside of bugzilla. 
which is the same now when a half-hearted buildroot is made
mandatory. If you want to make something mandatory it has to be
something worth doing so.

The buildroot that only covers a seldom corner case of multiple users
building the same package while ignoring the far more common use case
of building i386 and x86_64 on x86_64 (for F7 we're making even more
multilib developping noise) is just not worth putting in specfiles
lest to cast it into an iron mandatory part.

>  Let's move forward:
>  1) Make a clear rule on a buildroot value that fixes the technical
>  issues. 
We made that rule. BTW was it ratified? Still it will see opposition
just like the id -un rule. And any other buildroot is suggested.

So let us just stop suggesting a buildroot (or at least stop dictating
one). "If it works, it's OK". And any other funny corner case can
indeed bend buildroots at will, right?
-- 
Axel.Thimm at ATrpms.net

[Rex Dieter]

Axel Thimm wrote:


>  So let us just stop suggesting a buildroot (or at least stop
> dictating
>  one). "If it works, it's OK". 
Problem being there are even differing opinions on what "If it works" means.

-- Rex

[Axel Thimm]

On Tue, Feb 20, 2007 at 03:27:37PM -0600, Rex Dieter wrote:
>  Axel Thimm wrote:
>  
>  
>  > So let us just stop suggesting a buildroot (or at least stop dictating
>  > one). "If it works, it's OK".
>  
>  Problem being there are even differing opinions on what "If it works" means.
Let it be a broad as possible or drop that part from the
guidelines. "If it works" means it doesn't eat your cats, wife and
mother in law. Although some may even discuss that some of these
side-effects aren't really that bad ...
-- 
Axel.Thimm at ATrpms.net

[Axel Thimm]

spot asked me to draft something in the wiki about pushing all
responsibility to (grown-up) packagers while still presenting a couple
of sane buildroots as a guideline.

The outcome is on http://fedoraproject.org/wiki/PackagingDrafts/BuildRoot:

>  [[Anchor(BuildRoot)]]
>  == Build root tag ==
>  
>  The ''Build``Root'' MUST be below %{_tmppath} and MUST use %{name},
> %{version} and %{release}. It also may make use of ''mktemp'' since this
> is guaranteed to exist on any system. Other than that packagers are free to use any sane
> ''Build``Root''.
>  
>  The ''recommended'' values for the ''Build``Root'' tag
> are (in descending order of preference)
>  {{{
>  %(mktemp -ud %{_tmppath}/%{name}-%{version}-%{release}-XXXXXX)
>  %{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)
>  %{_tmppath}/%{name}-%{version}-%{release}-root
>  }}} 
At one point, this was a mandatory value, but it is now left to the packager.

On Tue, Feb 20, 2007 at 11:21:11AM +0100, Axel Thimm wrote:
>  So we voted on a new *mandatory* buildroot. By now people saw that
> the
>  previous *mandatory* buildroot entered the guidelines and started
>  blocking new packages requiring the old *mandatory* buildroot.
>  
>  I don't know what fesco did last week on ratifying or not the new
>  buildroot, and either way people will think differently on any single
>  buildroot. Perhaps buildroots are the most unimportant piece of s**t
>  with the most polarized parties.
>  
>  Now who's idea was it to have a *MANDATORY* buildroot at all?</rhetoric>
>  
>  Most of the FPC are fed up and have often stated that the buildroot
>  guidelines should be simple "If it works, have it".
>  
>  Plain and simple:
>  
>    Request for voting on dropping the *mandatory* from the guidelines
>    and explicitely cast it into a *suggestion*
>  
>  +1
>  
>  The first other five positive voters get a free beer when I meet
>  them (we never said that committee members could not be bribed, or do
>  we need a guideline for that? ;) 
-- 
Axel.Thimm at ATrpms.net

[Axel Thimm]

On Thu, Feb 22, 2007 at 09:21:38PM +0100, Axel Thimm wrote:
>  spot asked me to draft something in the wiki about pushing all
>  responsibility to (grown-up) packagers while still presenting a couple
>  of sane buildroots as a guideline.
>  
>  The outcome is on http://fedoraproject.org/wiki/PackagingDrafts/BuildRoot: 
Matthias Saou made some changes cleaning up my e-glish and clarifying
the one and other bit. In case you had made up your mind already and
don't want to feel like voting for something changing underneath your
feet I placed the changed content bits underneath the original
proposal, see also

http://fedoraproject.org/wiki/PackagingDrafts/BuildRoot?action=diff&r...

>  > [[Anchor(BuildRoot)]]
>  > == Build root tag ==
>  > 
>  > The ''Build``Root'' MUST be below %{_tmppath} and MUST use
>  > %{name}, %{version} and %{release}. It also may make use of
>  > ''mktemp'' since this is guaranteed to exist on any system. Other
>  > than that packagers are free to use any sane ''Build``Root''.
=> The ''!BuildRoot'' value MUST be below `%{_tmppath}/` and MUST
   contain at least `%{name}`, `%{version}` and `%{release}`. It may
   invoke `mktemp` since this is guaranteed to exist on every
   system. From there, packagers are expected to use a sane
   ''!BuildRoot''.

>  > The ''recommended'' values for the
> ''Build``Root'' tag are (in descending order of preference)
>  > {{{
>  > %(mktemp -ud %{_tmppath}/%{name}-%{version}-%{release}-XXXXXX)
>  > %{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)
>  > %{_tmppath}/%{name}-%{version}-%{release}-root
>  > }}}
>  
>  At one point, this was a mandatory value, but it is now left to the packager. 
=> At one point, the second was a mandatory value, but it is now left
   to the packager to decide. If unsure, simply pick the first.
-- 
Axel.Thimm at ATrpms.net