On Thursday, March 29, 2012 7:02:45 AM UTC+8, John H Palmieri wrote: > > > > On Wednesday, March 28, 2012 3:51:42 PM UTC-7, Jeroen Demeyer wrote: >> >> On 2012-03-28 17:08, leif wrote: >> > Simply making CVXOPT depend on matplotlib is certainly easier, but >> > IMHO not really appropriate since the former *builds* without the >> > latter. >> I don't see the problem. What's the harm with the easy solution? >> > > > Upgrading matplotlib would trigger a rebuild of cvxopt. In principle, if > this rebuild is unnecessary (that is, if SAGE_CHECK is not set), then it > could add a long time to the upgrade process. In practice, cvxopt is not a > dependency for anything else, and building cvxopt doesn't take very long, > so I think this is not an important concern. > Is there any other reason to worry about the easy solution? >
The question is: will anyone remember two years down the line why cvxopt had matplotlib as a dependency? Hypothetically, suppose 5y down the line someone else wrote a super-awesome new library that depended on cvxopt but which takes half hour to build. Then an update to matplotlib would trigger update to all of these three, and everyone will take it to be a normal state of affairs. Over time these complications will just accumulate until someday some exasperated release manager (Jeroen? ;)) decides to poke around deep into the history of all package dependencies. -- 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