Luca2 liliana.perossa at
Sat Apr 21 09:29:49 PDT 2007

----- Original Message ----- 
From: "Dan Nicholson" <dbn.lists at>
To: "LFS Developers Mailinglist" <lfs-dev at>
Sent: Saturday, April 21, 2007 5:03 PM
Subject: Re: Thoughts

> Who's not getting a chance to decide? Isn't the motto "Your distro,
> your rules"? All it says in the book is that these are tested
> instructions. The same thing applied when we were talking about
> filesystems. It says to format your disk as ext2, but that's just a
> known working method.

Hi Dan.

I didn't say that but simply that in current stable/svn books there are 
is mention (at least last time I read the books) about other 
possibilities (there are some hints).
Anyway you put under evidence a point: these are tested instructions, 
why not testing others?

I know the use of the "-B" option to solve the unsupported hash but my 
"argument" was another one: instead of putting the "-B" option why not 
testing newer binutils, glibc etc... ?
I know the statement of LFS book and the one Matthew quoted.

> I'm very interested in hearing about your experience with
> --hash-style, though. If you compile a binary with gnu HASH, you can't
> move it to another machine unless the glibc supports gnu HASH, right?
> --
> Dan

I had this problem, but really binutils must support it first; it's the 
only trouble I had trying to move binaries to another machine, the 
advantages I found are basically in a reduced dynamic linking time 
(about 50%) leading to a "faster" system; I tried gcc-4.3.0 too (and 
using it by default) without much troubles (I had only to patch a lot of 
glibc sources to compile it with gcc-4.3.0) and then these innovations 
with reiser4: the major bug I found in reiser4 is a system "stasis" 
after 10 working hours: a 10 minutes freeze.
About this one: why not adding a note in lfs book about the grub patch 
to support it?

Anyway, as already said, these are only my opinions, nothing else.


More information about the lfs-dev mailing list