static libraries' policy

[Christian.Iseli at licr.org]

veillard at redhat.com said:
> On Sat, Nov 12, 2005 at 08:21:28AM -0800, Toshio Kuratomi wrote:
> > If Daniel's got a valid argument then static libraries have to be
> > packaged by default.

>   At least for the libraries
>     1/ that we expect ISV to use
>     2/ that are not part of the LSB set
>  this affects me directly (libxml2) and potentially all desktop libs

I agree that applications *strictly* conforming to LSB will need to link 
statically to some libs.  But the LSB standards also says about conforming 
applications:
<quote>
It shall not use any interface or data format that is not required to be
provided by a conforming implementation, unless:

    * If such an interface or data format is supplied by another application
      through direct invocation of that application during execution, that
      application shall be in turn an LSB conforming application.

    * The use of that interface or data format, as well as its source, shall be
      identified in the documentation of the application. 
</quote>
and I'm pretty convinced that any package in FE satisfies those conditions 
when they are dynamically linked (just look at the BR lines in the spec, and 
the SRPMs of necessary libs are readily available).

I could not find any documentation that Fedora aims to be *strictly* 
LSB conformant.

There is also this old quote from Andreas Rogge in 2002:
<quote>
for strict conformance, it is required that the package does only
require LSB and not any other package.
This makes it impossible to make software suites (i.e. GNOME or KDE) LSB
conforming. As they consist of many applications, which require each
other.
</quote>
(from http://www.fedora.us/pipermail/fedora-devel.mbox/fedora-devel.mbox)

> > They don't need to be installed but they have to
> > be available.  Does moving libraries into -static for FE work?  Or are
> > we going to run into the same comps maintenence issues that Jeremy
> > mentions in relation to FC?

>   A no-nonsense rule is the best, we don't want to add hundreds of -static
> packages, a maintainer may have a good reason to ship static libs, LSB being
> one of them, apparently Chris Aillon gave another example where it's required
> for Mozilla. No static should be a general trend, and for a number of
> libraries it makes sense, but this should not be imposed as an absolute hard
> rule. 

I agree that a no-nonsense rule would be good.  I still stand by my conviction 
that static libs should be removed by default.  But I have no problem granting 
people wanting their packages to be strictly LSB conformant their wish to pack
the static libs.  I just think they should have given the issues of:
 - security
 - bulk
 - maintainability
some thought, and have some justification before deciding to pack the static
libs.

					Christian