On 9/11/18 7:36 AM, Bruce Dubbs wrote:
On 09/10/2018 11:10 PM, Kevin Buckley wrote:
I have no problem with you making those changes, but will raise the
issue of
whether the dump-commands is useful at all. Why not just remove the
dump-commands target from the Makefile?
But it seems to do exactly what its name suggests it should,
if run as part of the "all:" target.
Indeed, I have been comparing it's output with the commands
I ran for an 8.0 LFS build, that I had saved by hand (so as to
add extra XML into my local copy of the 8.3 Book), and, give
or take a few expected differences, it's a match.
If anything the "dump-commands" target is a useful test of whether
the "examples" and "real commands" are marked up correctly, no ?
jhalfs does that. Check it out.
Well, not in chapter 4 ;) Right now, in jhalfs, the commands from
chapter 4 are not read from the book.
It's not easy to automate chapter 4 commands, because they depend on the
existence of the lfs user and specially the lfs home directory and the
files in it. If there is already an lfs user, and for some reason, the
bash startup files are different from what is in the book, it may not be
fair to change them... One way out of this could be to always create a
new user (with a random name, say x78tu6), but then, all the commands
such as "groupadd lfs" should be rewritten by jhalfs to "groupadd
x78tu6". Not sure that the equivalent (in xsl language) of a sed command
"s/lfs/x78tu6/" would do.
Another possibility could be to run chapter 5 as $USER (not lfs), and to
use an "envar" file source'd (possibly with environment reset) at the
beginning of each scriptlet. That would depart from the book, though.
Pierre
--
http://lists.linuxfromscratch.org/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
