r8518 - in trunk/BOOK: . chapter01 prologue

[Dan Nicholson]

On Fri, Apr 11, 2008 at 3:40 PM, Sukucorp Sukucorp <sukucorp at gmail.com> wrote:
> On Fri, Apr 11, 2008 at 4:30 PM, Dan Nicholson <dbn.lists at gmail.com> wrote:
>  >
>  >  yacc (bison) - Definitely needed since we patch the bash parse.y file
>  >  in Ch. 5 and yacc will need to be rerun. Could add bison to Ch. 5
>  >  before bash, but IMO it's easier for the host to just install bison.
>
>  I don't remember if bash calls bison or yacc. If it is the former, we
>  don't need to check yacc -> bison symlink.

If it uses the autoconf macro AC_PROG_YACC (which I would be very,
very surprised if it didn't), then the search goes: bison -y, byacc,
yacc. So, I suppose any yacc would be fine, although I'd be surprised
to find any Linux system where this is not bison.

>  >  awk - Definitely needed by the glibc build, and gawk in particular due
>  >  to use of some non-portable features.
>
>  glibc needs gawk but it doesn't need awk -> symlink.

True. Same thing as above with AC_PROG_AWK: gawk, mawk, nawk, then awk.

>  >  bash - Not needed for actual building, but glibc's ldd and tzselect
>  >  need either bash or ksh to work. The values will be substituted at
>  >  configure time. I don't know what happens without them, and it's
>  >  probably not that important in Ch. 5 if those utilities aren't there.
>  >  However, we create the LFS user with /bin/bash as the login shell, and
>  >  this can't be substituted as is because we set up the environment
>  >  through the bash initialization files.
>
>  Ditto. bash has to be installed but need not be the default shell.

Agreed (although there are plenty of shady configure scripts in the
wild that use non-Bourne or non-POSIX syntax).

>  Randy echoed my thoughts in a much better manner. We should not check
>  for something if the result of the check does not make any difference
>  in the build.

Agreed, and I don't think we're doing that right now.

--
Dan