Hi Georg, On Sun, Jan 24, 2010 at 10:06 AM, Georg S. Weber <georgswe...@googlemail.com> wrote: > Hi, > > I just noticed the following. In the Sage-4.3.1 source tarball, there > are both the packages "lapack-20071123.p0.spkg" as well as > "lapack-20071123.p1.spkg".
Indeed there are. > If you build Sage from source, no problem > (yet), under /spkg/installed/ there will be only the newer "p1" spkg. > But if you "sage -bdist ..." a binary Sage distribution, then, under / > spkg/standard/, for both the "p0" and the "p1" spkg, a placeholder > file will be created (that shouldn't happen). Confirmed. > Worse, if I take this binary distribution, install Sage (elsewhere) > from that one, and issue "make test", then happily the "p0" lapack > spkg will not be found as installed (it isn't, "p1" is installed --- > but the "p0" shouldn't be looked for, which is bad), and then Sage > begins to download the "p0" one from the web and begins to build it > (deleting the files from the "p1" spkg first, according to the logs). > That should not happen at all! Confirmed. > I hardly got accustomed to the fact that "make test" in the above > situation will, at the beginning, output quite some crap, e.g. about > "circular dependencies dropped", and then rebuild just all of the > documentation files (taking a long time), for nothing --- these were > already provided. Ticket #7943 [1] tracks this problem. > Did anyone experience the same problem(s)? We should create some > tickets, then. This is now ticket #8047 [2]. Like the recent problem with Sage 4.3.1.alpha0 [3], a fix here is to delete lapack-20071123.p0.spkg from SAGE_ROOT/spkg/standard (in the source tarball) and leave lapack-20071123.p1.spkg alone. [1] http://trac.sagemath.org/sage_trac/ticket/7943 [2] http://trac.sagemath.org/sage_trac/ticket/8047 [3] http://groups.google.com/group/sage-devel/msg/08cf0bd4afb3cf01 -- Regards Minh Van Nguyen -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
