r8405 - in branches/jh/BOOK: chapter01 chapter05 chapter06

Ken Moffat ken at linuxfromscratch.org
Sun Oct 7 17:04:39 PDT 2007


[ 2nd attempt at replying - an explanatory note just in case the
first reply from my ntl address does eventually get through ]

On Sat, Oct 06, 2007 at 08:14:46PM -0600, Jeremy Huntwork wrote:
> On Sat, Oct 06, 2007 at 08:09:45PM -0600, jhuntwork at linuxfromscratch.org wrote:
> > Author: jhuntwork
> > Date: 2007-10-06 20:09:45 -0600 (Sat, 06 Oct 2007)
> > New Revision: 8405
[...]
> > Log:
> > Made book buildable by powerpc.
> 
> It took a full day to build on my G4, but with these two small changes
> jhalfs ran smoothly and completed successfully. Now that I have the
> basics ironed out, it's time to let one run with the full teststuites
> enabled, and possibly ICA/farce.
> 
> Ken, care to make a test run?
 Sure, I was hoping to get back to testing _something_ this week, so
why not ppc. [ OTOH, last week I expected to build a desktop with
current packages, and didn't really get beyond firefox - at the
moment my estimates are very poor in time terms, at the moment there
are other things more important or more fun than building software. ]

 For testsuites, expect 1 additional failure in e2fsprogs - I assume
it's a big-endian-specific failure (I forget which, but google
suggested the test was against a big-endian corner case, and
research showed the failures started in 1.40-WIP versions).  Despite
the failure, I'm reasonably confident about using 1.40 on ppc (in
other words, yet another case where I find the test suite probably
unhelpful).  Of course, if it passes all tests for you I'll have to
revise my views.

 As to farce, I've pretty much abandoned it - 6.3-rc on x86 had too
many not-adequately-explained differences across multiple builds
(particularly, files that sometimes matched and other times
differed).  Bit-for-bit repeatability is nice if you can get it, but
I'm no longer sure that its absence is necessarily a problem (there
were always a few files that differed, e.g. the undocumented locales
archive, but the number seems to have grown recently.

ĸen
-- 
das eine Mal als Tragödie, das andere Mal als Farce



More information about the lfs-dev mailing list