Re: apparent serious integrity problem in build system - setuptools bug?

2014-03-21 Thread Leif Ryge
On Fri, Mar 21, 2014 at 08:01:48PM -0400, Greg Troxel wrote: > > Update: after spiffing up py-OpenSSL quite a bit more, so it passes its > own tests, tahoe seems to work ok with it. I'm still not sure, and > will check further. > > My comments about auto-downloading stuff by default being a bug

Re: apparent serious integrity problem in build system - setuptools bug?

2014-03-21 Thread Greg Troxel
Update: after spiffing up py-OpenSSL quite a bit more, so it passes its own tests, tahoe seems to work ok with it. I'm still not sure, and will check further. My comments about auto-downloading stuff by default being a bug stand :-) pgpRsCiBIrlFL.pgp Description: PGP signature

Re: apparent serious integrity problem in build system - setuptools bug?

2014-03-21 Thread jg71
* jg71 wrote: > Best case would be (and let's be optimistic here) that tahoe just > works with pyOpenSSL 0.14, including the new deps on six and > cryptography. > > Anyway, I'll get back to you after having run some tests. After having dealt with the new pyOpenSSL situation, running tahoe -V

Re: apparent serious integrity problem in build system - setuptools bug?

2014-03-21 Thread jg71
* Greg Troxel wrote: > Zooko Wilcox-OHearn writes: > > > To change it so that it *doesn't* automatically download packages > > #over the Net is ticket 1220. > > OK - this is the basic bug. By default, it should fail with an error, > and do this autofetching if given the --autofetch or whateve

Re: apparent serious integrity problem in build system - setuptools bug?

2014-03-21 Thread Greg Troxel
Zooko Wilcox-OHearn writes: > 1. The Tahoe-LAFS build system (which is based on setuptools) will > download and execute any package which pypi.python.org says is the > "foo" package, if any part of Tahoe-LAFS transitively depends on a > package named "foo". To change it so that it *doesn't* auto