In message <20051102161314.22DA75283@fnord.ir.bbn.com>, Greg Troxel writes:
>Since upgrading from current of 20050202 to 20050926, backups have
>been failing.  The problem is in revision 1.73 of src/usr.bin/gzip.c,
>where if the output size is > 4GB then a warning is printed.  But,
>maybe_warn causes exit status to be 1.  It seems that this isn't
>really an error condition, and that at a minimum exit status should
>still be zero.

Indeed.  In fact, that was my solution; maybe_warn doesn't change exit status
on my system, so Amanda warns me about the possible confusion (which is good),
but doesn't FAIL.

>Plus, this should be maybe_warnx, not maybe_warn, because errno has
>not just been set by any library call and there's a stray EOPNOTSUPP
>that doens't make sense.

I don't think so; doesn't the 'x' imply exiting or something?

>I'm going to just delete the warning line in my local sources since
>backups are more important to me than being told the length will be
>%4GB.  Does GNU gzip output a warning?  return 1?  I'll ask on the
>amanda list to see if anyone else has this trouble.

I think GNU gzip just silently produces archives it can't test.  Which
is bad, but not as bad as causing scripts to abort and report spurious errors.

-s