Matthew Burgess wrote:

> What I have trouble understanding is the fact that, apparently, one
> shouldn't reboot during the ICA cycle.  What I thought was trying to be
> proved here was that a) any suitable host can build LFS

No. IMHO what you refer to there is "bootstrapability" ie: the ability to
bootstrap a Linux system from an existing Linux host system. ICA measures
something different ie: "reproducibility" (I've also seen it referred to
by some BSD folks as "repeatability").

> and b) regardless
> of the host, the final LFS system should be more-or-less identical
> (ignoring timestamps embedded in files, etc.).  Therefore my initial
> *assumption* would have been to do something along the lines of:
> 
> 1. Build LFS from any suitable host
> 2. *Reboot* into that LFS system
> 3. Build LFS from the LFS system built from stage 1. 4. Compare the
> systems built in step 1 and 3.

No. That won't catch what ICA is trying to detect. Let me try to explain
it. This gets back to the very essence of the build method:

Any source package has a basic right to be configured and built and in a
sane environment. Early Chapter 6 is *not* a sane environment. The
completed Chapter 6 (with /tools removed) *is* a sane environment (yes,
this is an assumption). The whole 2 phase build method approach is
designed to try and provide as sane an environment as possible. The goal
is to produce an end result identical to that which would have been
produced as if you had produced it in a sane environment. This is what ICA
can verify!

Here are the basic steps to ICA:

 - build Chapter 5 (as normal)
 - build Chapter 6 (as normal)
 - take a snapshot
 - remove /tools from the PATH
 - use Chapter 6 to rebuild itself (overwriting everything as it goes)
 - take another snapshot
 - AGAIN use Chapter 6 to rebuild itself (for good measure)
 - take yet a third snapshot
 - now binary diff the snapshots (1 versus 2, 2 versus 3)
 - explain any differences
 - any unexplainable differences are BUGS in the build method.

See? ICA does not guarantee a good build. It's just a simple measure of
the effectiveness of the 2 phase build method approach. This is why those
folks tinkering with the build order should verify their results.

Regards
Greg
-- 
http://www.diy-linux.org/

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

Reply via email to