Randy McMurchy wrote:
However, LFS history has shown that we cannot count on such a document
to become formalized.
I'm not sure if a formal set of rules is in fact possible. If we
consider the packages that are in the book at the moment, they can be
broken down into roughly these areas:
1. Toolchain
2. Required/useful optional build dependency (e.g. ncurses, zlib, readline)
2. Useful development tools (autotools)
3. FHS/LSB stipulated binary (coreutils, grep, gawk, etc.)
4. Historical/accidental (psmisc, procps, etc.)
So, now try converting that into a formal specification!
Obviously the above list is useful in terms of stopping packages like
Apache and MySQL getting into LFS, but it seems to me there's enough
room to manouevre using that list that if someone proposes a new package
it will more than likely lead to a discussion such as this one. The
benefit of having a formal 'Package Criteria Specification' document is
therefore debatable, IMO.
Regards,
Matt.
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
