Hi Tom, I had thought I looked at this, but perhaps I forgot? Apologies if that's the case.
On Fri, Jul 17, 2009 at 09:44:04PM -0500, Tom Mueller wrote: > Still looking for a review for this. Also, I added one additional bug > related to running on Windows: > > 10127 transport module uses os.statvfs which isn't available on Windows > > Thanks. > Tom > > > Tom Mueller (pkg-discuss) wrote: >> Please review this changeset for making the Windows build and unit >> tests work with the recent pycurl additions and other changes. The >> issues are: >> >> 8656 build download hash check fails in Windows due to text mode open >> 9997 transport.engine uses SIGPIPE which doesn't exist on Windows >> 9998 URL construction incorrect in p5i and t_p5i >> 9999 t_p5i test case leaves file descriptor open >> 10000 pycurl and CA certs fail to build on Windows >> 10001 t_plat test cases fail because working directory is inside image >> to be removed >> >> The webrev is here: >> http://cr.opensolaris.org/~tmueller/ips-winbuild/ >> >> These changes are implemented according to the suggested fixes in the >> bug reports. >> >> Thanks. >> Tom >> >> P.S. Please ignore the links in the webrev index page for the bug ids >> 10000 and greater. This generally looks fine to me. The only request I have is that if we _know_ that we're on a Windows system on line 396/397 of transport.py, then can we set the filesystem blocksize to the default blocksize for Windows filesystems? The engine picks 128k by default, since that works with ZFS. This check is just here to make sure that some other blocksize hasn't been selected, and if it has, adjust the buffering appropriately. Thanks, -j _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
