On 07/ 3/10 05:49 AM, William Stein wrote:
Hi,
I still vote -1 to this, and think it is possible to get around using
patch at runtime.
Nonetheless, I am ok with this proposal going forward, because it
clearly received a lot of support from most developers who commented.
William
OK
http://trac.sagemath.org/sage_trac/ticket/9418
is open to create the package with a target of Sage 5.0.
There are lots of tickets open now for updating spkg/standard/deps, so these
need to be coordinated. Actually adding the package will be pretty easy then.
One question I do ask, is whether the self-tests of the package should be run
automatically. If 'patch' fails to work properly, people are going to have very
broken builds of Sage. The time to install the package increased from 3.2 to 3.7
seconds on my Ultra 27, so I personally think adding 0.5 seconds to the install
time, and testing the 'patch' package is a good idea. In other words, we test it
on every installation, not just if someone has SAGE_CHECK=yes.
The package at
http://boxen.math.washington.edu/home/kirkby/patches/patch-2.6.1.spkg
follows the conventional method of having a spkg-check file, and running the
self-tests only if SAGE_CHECK=yes. I think it would be wise that the package
runs the self-tests.
One problem of running the self-tests is that it would probably require 'diff'
exist on the system. I would have thought 'diff' existed on any system. It is
mandated by POSIX, but I've got no way of knowing if every version of Linux
comes with 'diff' or not.
Another related item is how to update the Sage documentation to reflect how to
make the patches. We need to tie down the exact method and naming convention for
patches.
http://trac.sagemath.org/sage_trac/ticket/9419
has been opened to update the Sage Developers Guide, once agreement is reached
on this. I will create another thread on sage-devel on how to actually create
the patch files as different people might well have different ideas on how to do
this. Let's keep that a separate thread.
Dave
--
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
