On Sat, Jul 15, 2006 at 05:51:22PM -0400, Joe Ciccone wrote:
> Alex Merry wrote:
> > Robert was suggesting a change that would allow the sed to be applied to
> > other *FLAGS variables. When you consider the dangers in constructing and
> > running commands you don't fully understand in the compilations of
> > toolchain packages, it seems to me that it would be sensible to actively
> > discourage such an activity.
> >   
> No, His suggested sed would still only affect that one line. I'll even
> copy it from his origional email so you can look at it again.
> 
> sed 's/^XCFLAGS =/& -fomit-frame-pointer/'

Sorry, I wasn't very clear about what I said. The point I was trying to
make was not that his change would affect other *FLAGS, but that the
_rationale_ for his change was that it would make it easier to adapt to
other *FLAGS variables.

> > In fact, I believe Robert misunderstood the sed, thinking it was
> > something like 's/^XCFLAGS =.*$/& -fomit-frame-pointer/'. If they tried
> > to adapt, say, XFOOFLAGS using this method and XFOOFLAGS looked like:
> > XFOOFLAGS = -blah -foo \
> >   -bar
> > then the sed 's/^XFOOFLAGS =$/& -baz/' would do precisely nothing. So
> > users trying this method who don't know what they're doing end up doing
> > nothing rather than screwing up their compilation.
> >   
> Let's quote something from the lfs book. In section 1.5.1 there is a
> note that states exactly the following, " Deviating from this book does
> /not/ mean that we will not help you. After all, LFS is about personal
> preference. Being upfront about any changes to the established procedure
> helps us evaluate and determine possible causes of your problem."
> 
> There is no reason for this sed to have the $ in it. I may apply a patch
> to gcc that adds a value to that line. This is deviation from the book,
> but in fact, the sed with the $ in it will render the command useless if
> I update this line myself. Why make it hard for people who want to
> deviate from the book when in section 1.5.1 it says that lfs is about
> personal preference. Maybe my personal preference is to add something to
> the XCFLAGS line before I run the sed but, I overlook what the sed does.
> It just simply doesn't make sense to me to change the command to benefit
> others without sacrificing anything!

Fair enough. I wasn't convinced by Robert's reasoning, partly as it
seemed to be based on a misreading of the original sed command and
partly because the idea of adapting sed commands you don't understand
makes me very wary (been there, done that, shredded the T-shirt), but
this seems like a perfectly good reason to change the sed, assuming there
are no other lines in gcc/Makefile.in that start with "XCFLAGS =".

Alex :-)

-- 
Pippin
Computer Monkey to the Pelican
www.oxrev.org.uk, www.corpusjcr.org, www.rev.org.uk

Attachment: pgpAZMTfuXlPN.pgp
Description: PGP signature

-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Reply via email to