Strange GCC failures in Chapter 5

Greg Schafer gschafer at
Sun May 11 21:08:40 PDT 2003

On Sun, May 11, 2003 at 11:39:04PM -0400, Zack Winkles wrote:
> During the make check of GCC pass 2 in Chapter 5, all of the usual
> XPASS's in libstdc++-v3 become FAIL's. Though I don't have NLS on the
> host, it shouldn't particularly matter should it? It should be using the
> Glibc that we built in Chapter 5, and therefore it's locale data. I
> checked all of the flags, all of the Chapter 5 Glibc's locale stuff,
> everything. It just seems strange.

Works for me. Cannot guess as to what went wrong for you.

> Despite this, GCC in Chapter 6 has results right along the lines of what
> we would expect. 26 XPASS's and 1 FAIL. So it seems whatever was/is
> wrong doesn't get passed on to our child LFS system.
> Due to this, wouldn't it make sense to remove the checking of GCC (and
> maybe all the other packages, since their results don't get passed on to
> the child)? It's wasted effort it seems to me. PLFS does it's job, and
> the LFS is pure. Out with Chapter 5 checking!

I had many a long dialog with Ryan about this back in the beginning. There
are pro's and con's for both scenarios and the hint tries to present the

IMHO, the book should present the facts to the user and let him/her decide
for himself/herself whether to run the Ch 5 test suites or not.

I case you're wondering, Ryan was always more anal than me so you can guess
which way I'm leaning :-)

Unsubscribe: send email to listar at
and put 'unsubscribe lfs-dev' in the subject header of the message

More information about the lfs-dev mailing list