Thanks for this post, Bruce.
There are two disks sda and sdb in my system, with partitions sda1 for LFS7.0
and sdb1 for Ubuntu 11.10. This is to justify why it will take some time until
I will give you a return about the install script you kindly mailed me.
That said, I will address the other points.
Em 16-12-2011 00:31, Bruce Dubbs escreveu:
> Fernando de Oliveira wrote:
>
>>> What it says is "Compilation OK", not that all packages are OK. It's
>>> supposed to be a short script to allow the user to visually check the
>>> actual installed packages. That's part of the learning process.
>> Yes, but is it really difficult to drop "Compilation OK" and add
>> "Host System Requirements OK/failed" ?
> That would require a lot longer script. Every line would need to be
> parsed to determine the version and then checked. Since every line is
> slightly different, it becomes a much more complicated script.
Agreed. I was thinking about some flag that could be used just to write, after
"Compilation OK", to state that some requirements should be fixed, "please,
read carefully the output above", if the flag, initially 0, had been modified.
If not very difficult, the final value of the flag would be the number of
missing requirements, which would improve the message to "$flag required
packages have to be fixed".
>
> On the other hand, each line in the script tells the user how to
> determine the version. That information is valuable in itself.
That is *very* valuable information.
I need to say, this has not been a problem for me since perhaps LFS6.5. I am
only insisting in it, because I remember that it took me a very long time to
sort this out, and I believe a lot of newcomers will give up after something
like that.
I feel it is misleading having a statement "compilation OK" and at the same
time "some requirements not satisfied". Perhaps, instead of any major changes
in the script, a message could be given (or the previous messages could be
changed) to avoid the understanding that "although some requirements are
missing, the script is telling me that compilation is OK, so I do not have to
worry for those previous warnings, because the system will be able to compile
the packages". What I mean is: perhaps a small modification would avoid
misunderstandings in something that tells OK and missing. simultaneously.
>
>>> Perhaps a comment. What is the output when you run
>>> /lib/i386-linux-gnu/libc-2.13.so as an executable?
>> lfs@VMWLFS70:~$ /lib/i386-linux-gnu/libc-2.13.so
>> GNU C Library (Ubuntu EGLIBC 2.13-20ubuntu5) stable release version 2.13,
> Ahh, EGLIBC. I believe this is supposed to be binary compatible with
> glibc. Perhaps the following would work in the script:
>
> find /lib -name 'libc-*.so' -exec {} \; | head -n1 | cut -d, -f1
lfs@VMWLFS70:~$ find /lib -name 'libc-*.so' -exec {} \; | head -n1 | cut -d, -f1
GNU C Library (Ubuntu EGLIBC 2.13-20ubuntu5) stable release version 2.13
If I understand correctly your point, it works great, and some comment in the
page will address how to make the symlink /lib/libc.so.6 -->
/lib/i386-linux-gnu/libc-2.13.so, as compilation fails without that.
--
[]s,
Fernando
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
