Minimum Host Prerequisites
jhuntwork at linuxfromscratch.org
Mon Oct 20 08:26:12 PDT 2008
Ken Moffat wrote:
> On Mon, Oct 20, 2008 at 10:36:14AM -0400, Jeremy Huntwork wrote:
>> * "Gcc-3.0.1" can at _least_ become "Gcc-2.95" I don't know if you
>> want to mention "egcs-2.91.66" but it works.
> Possibly, but if you have to build a linux-2.6 kernel on the old
> host, you won't be able to build a recent version of 2.6 with those
> old compilers.
> In an *ideal* world, I still think that building the kernel version
> used in LFS on the host is the way to go. Doesn't fit every usage,
> but it has to be a lot better than attempting to support people
> building from some antique version of 2.6. Iff people are able to
> build successfully with kernels < 2.6.16.recent [ chosen because
> 2.6.16 has long-term support ], that's good, but I don't think we
> should be going out of our way to encourage it.
Well, technically, you should be able to build a kernel with your first
pass of gcc, load it into your bootloader, reboot and pick up where you
left off. I haven't gotten that to work yet, but if I do, I'll let you know.
> FWIW, my _first_ experience with linux was using redhat 6.2 - that
> was with a free CD from a magazine in late 1999. Things have moved
> on. For the desktop, I can be rightly accused of not liking certain
> modern developments, but even I think 9-year-old software is not a
> normal place from which to start.
Nope, definitely not normal. And I wasn't suggesting that we recommend
it to anyone. My goal in doing this was to get a sense of just how
robust LFS currently is. If it can build itself perfectly from such an
old and crusty system, that's saying something.
The 6.2 host is very fragile. If something is 'off' it lets you know
right away. I rather like that... :)
More information about the lfs-dev