On Fri, Sep 12, 2014 at 04:15:09PM -0500, Bruce Dubbs wrote:
> Fernando de Oliveira wrote:
> >On 12-09-2014 17:32, Fernando de Oliveira wrote:
> >
> >>
> >>Starting from scratch.
> >>
> >>Thanks for the help, Bruce.
> >>
> >
> >Forgot to say: disabled all tests in jhalfs build, they are worthless.
> 
> With -j1 they are good for editors to check the instructions when new
> package versions are introduced.  For newbies they are good to check that
> things are going correctly.
> 
[ long response, skip unless you really care about testing - in that
case, you probably remember the strange results some people had in
the LFS-4 variants, and like me welcomed the move to "pure LFS" with
the tests. ]

 Not worthless, but not usually worthwhile if you have no recent
results to compare them to!  In _extreme_ cases they DO show
problems (in what is now the distant past, when Ryan was still
around and I was building on my not-missed AmigaOne (ppc) as part of
the test team, they showed a problem - in the end, that CPU died
because it overheaded one time too many), and for LFS editors they
are indeed useful.  Similarly, in the early days of cross-lfs : Jim
moved something, the findutils tests started to fail.

 Also, my own scripts use 'make check' without $MAKEFLAGS, except
for one of the autotools packages which I think you changed - I have
never otherwise used -j2...-j4 for check, nor for install, it seems
too risky.

 So, I hope that there will be results to compare to, by the time
7.6 is released -  The link to the gnu lists did not immediately find
any _relevant_ results (for x86_64-linux only results from the
development 4.9.2 and 5.0 branches).

 Also, you all know that I am prone to making mistakes in my
scripts, particularly by missing things that changed.

 For the moment, I have not looked at my i686 test results (only
finished the desktop build a few hours ago, and when I'm at the
machine my priority is to test the applications, including those
not in BLFS (they look good so far).

 I'm not proposing to upload my own results, both because of these
failures which _might_ be down to local errors, and because I build
in /building for chapter 6 - the old "my /sources is on nfs"
difference - we seem to be attracting increasingly inexperienced
users, I fear that the references to /building would confuse them.
Oh, and I forgot to run tests for a couple of packages which used to
be in BLFS, and for i686 I did not build everything (acl, possibly
attr, and I also forced the CFLAGS to -O2 in i686).  As always, I
can upload these if it would be useful.

 So for the moment I will merely note that on x86_64 I had 3 perl
test failures (something like 99.8% passed, I think, but more than
0.1% failed, so not "good"), and some torture failures in gcc.  I
have seen both of these in svn, and have no expectation that they
matter (both my August-svn and 7.6 builds have _built_ almost
everything I normally build, and so far I have not seen errors
attributable to these test failures).

ĸen
-- 
Nanny Ogg usually went to bed early. After all, she was an old lady.
Sometimes she went to bed as early as 6 a.m.
-- 
http://lists.linuxfromscratch.org/listinfo/blfs-dev
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Reply via email to