Damir Shayhutdinov пишет:
> 
> 1, Explicit linking dynamic objects with all dso it needs ensures
> correct soname dependencies and prevent ABI incompatibilities.
> 
> 2. Unused direct dependencies encumber package with extra dependencies
> (including soname dependencies). An encumbered program takes more time
> to load (5% in case of gdm for example), and need more memory mapped
> to its address space.
> 
> An encumbered package requires extra packages when installed, and
> needs extra care when one of unused libraries was updated and its
> soname was changed.
Увы. Переговоры с апстримом закончились ничем... Править мейкфайлы
для поддержки --as-needed автор отказался. Аргументы оказались
неубедительными.
David> A linker feature that makes `make install' of my program run
David> 5x slower, makes module loading 50% slower, etc. is not
David> something that worths to support.
David> In any case, --as-needed is a hack.  It subverts the normal
David> linker behaviour -- based on certain assumptions that may or
David> may not be true.  They are not true for Gwyddion.  Granted,
David> the linking scheme used by Gwyddion is not common, but
David> there's nothing incorrect on it (in fact, it prevents some
David> disasters that can occur when modules are linked with an
David> independent set of libraries).  The correct approach to
David> --as-needed would be probably to detect it's in use in
David> configure and add --no-as-needed to linker flags. This way
David> everyone would be happy -- execept perhaps people who made
David> it a religion.
и т.д...
А т.к. это, похоже, исключительно наша фишка, нужны боле сильные
аргументы или правка в процессе сборки rpm-а.
Насколько я знаю, rpm не умеет вызывать при наложении патчей
autoconf/automake. Так? Или я не прав?
-- 
Rgrds,
Andriy
*********************************************************************
email: dobr at iop dot kiev dot ua            Kyiv, Ukraine
Phone: (380-44)   525-7824            Department of Gas Electronics
Fax:   (380-44)   525-2329             Institute of Physics of NASU
*********************dobrATjabber.iop.kiev.ua************************