Current Ch 5 Commands

Ryan.Oliver at Ryan.Oliver at
Mon May 19 21:35:23 PDT 2003

Greg Schafer wrote:
> Ch 5 - Expect - I think all the "-with-tcl=/stage1/lib --with-x=no
>  --disable-symbols --cache-file=/dev/null" stuff came from Ryan's
> AFAICT, it doesn't achieve much and works fine without. We can drop it
> reason mentioned above i.e. less typing blah..

Will depend a lot on whether x and tcl are on the host system you are
coming from.
If you can guarantee it will pick up the proper tcl libs (and it should
after we have locked in our glibc) drop that option definitely, but please
don't drop the "--with-x=no" or else if you have X it will find it (and
cause you untold trauma in ch6 when it can't find the X libs) as it
specifically looks in /usr/X11R6 NOT PATH.

The original options specified work all the time for most every build host,
as everything is defined explicitly. YMMV when dropping them...

> Ch 5 - Ncurses - Same as mmap_test patch above. etip patch is not
> here. AFAICT, it is part of the ncurses c++ binding which is utterly
> in Ch 5. Anybody who needs the ncurses c++ binding in Ch 5 will know that
> they need it and can thus use it. But for the general book case, lets
> the patch here. Again, leave in for Ch 6. Same goes for the
> It apparently fixes some warnings in the ncurses c++ binding? Do we
> care in Ch 5? I think not (in fact I'm not sure we care in Ch 6 either -
> C++ expert should comment here then we can decide if the patch is
> appropriate for the book in general).

Depends if you like building stuff with deprecated headers or not I guess
(/me shudders). If we are building the c++ binding, build it properly.
Otherwise, drop the c++ binding entirely.
My $0.02

> Ch 5 - Bash - we can drop "--with-curses" now as the new build method
> ensures we only link against the /stage1 stuff where we only have


Unsubscribe: send email to listar at
and put 'unsubscribe lfs-dev' in the subject header of the message

More information about the lfs-dev mailing list