RFC: Is there a use for blfs-tool?
pierre.labastie at neuf.fr
Sun Feb 19 10:42:07 PST 2012
Le 18/02/2012 23:23, Thomas Pegg a écrit :
> I personally have never seen much use for blfs-tool. I've thought about removing it before too. It's hard to automate BLFS in the same way as we do LFS or CLFS, being it's not linear. While I believe it does try to do dependency checks, its not easy to code or maybe it is, I've never really looked to deep into it. I just know in comparing it to say gentoo and their portage tool, that the code inside portage is really complex (in Python no less). I personally have no objection to removing it, but there may be those that want to keep it.
Note that I do not want to remove (yet) the BLFS directory
and the tools it contains. Just the blfs-tool script.
Actually, I am finalizing a tool which should allow
automating much of BLFS, including Perl modules,
python modules and D-Bus bindings. I succeeded
in detecting circular dependencies and in linearizing
the book according to user instructions (choice
of which package you build first when a circular
dependency is detected), and actually everything is
working, except I am having somme trouble with the dependencies
in BLFS, which are not always up to date. Please do not consider what I
just said as a critic. Reading the blfs-dev mailing list,
I understand how hard it is to maintain this book. I am actually
offering to help with this tool, which, being dumb as a
tool, unveil many deficiencies...
For example, building an X server in the order of the chapter
24 works, but trying to build by saying "I want xorg-server,
so I go to the dependencies of xorg-server, then to the
dependencies of them and so on, until I find packages with no
dependencies, which I build first and then I go back building
packages once all their dependencies have been built", it ends up
with a wrong order (and actually no mesalib!). Well, becoming OT,
More to come in a forthcoming thread.
More information about the alfs-discuss