proposal: new approach

Tobias Gasser lfs at
Fri Jul 24 13:51:25 PDT 2009

as the current BLFS is quite outdated, i propose another approach to get
 the book up to date.

i guess there are to many outdated packages which are dependencies for
others. fixing one package requires to rebuild another bunch of packages
where one or another can't be built as it depends on some more packages
which themselfs require to update another one...

in my humble opinion updating the complete blfs-book is almost impossible.

*just start with a new EMPTY branch and RESTART FROM SCRATCH!!*

first get the most common libraries (like berkeley-db, jpeg, png ...) up
to a current version.

next would be x, gnome, kde and the most used servers like apache,
samba, ldap... just bare without anything on top of it.

and now the packages which some of the authors/contributers use
themselfs on top of kde / gnome or as an extension to one of the servers
already built.

but just step by step. example: just minimal php, no additional packages
required which aren't built yet. getting a full blown php requires to
much additional packages, but getting a minimal php is quite easy.

i guess with this approach we will have a blfs 6.x very quick. not as
complete as the current one, but the packages in the "new" book are well
tested and fit to each other.

whenever i wrote about a successull update of one of these packages
(last was berkeley db) i get a similar replay: the package has some more
dependencies which are not tested yet, thanks for the info, but no
update possible now.

i know i might have overseen some issues, but i'm really frustrated as
every hint i put back to the community is not applicable due to some
reasons which lead to a no-go for my commitment. everytime the arguments
are quite reasonable and i can understand my input can't be implemented
without any side effects.

as i could see during the last months, almost every input from a user
was rejected with "thanks for your input, but you didn't test package x,
y and z".

i myself have updated almost all packages i use from blfs (the few i
didn't had no update since the last complete book!!). but i have not the
complexity in dependencies as given by the COMPLETE book.

when i started with LFS there was no real BLFS but a long list of really
usefull hints.

now we have a BLFS which is outdated and hints which are outdated too.

for me a stipped down BLFS wich is current, and hints which are based on
a book (hints should correspond to a given book version!!) would be much
more desireable than the current status.

i don't want to offend anybody, and i really appreciate the work all the
authors are doeing. but the current state of the book is just [censored]

thanks for anyone haveing read 'til here!!


More information about the blfs-dev mailing list