Leo Famulari <leo@famulari.name> writes:

> On Mon, Aug 15, 2016 at 03:01:00PM +0100, Marius Bakke wrote:
>> 
>> > From c95b25a3ad4902ccdef79c7429485a7cacc72e1c Mon Sep 17 00:00:00 2001
>> > From: Marius Bakke <mbakke@fastmail.com>
>> > Date: Sun, 14 Aug 2016 16:47:33 +0100
>> > Subject: [PATCH] gnu: Add python-reportlab.
>> >
>> > * gnu/packages/python.scm (python-reportlab, python2-reportlab): New
>> >   variables.
>> > ---
>> >  gnu/packages/python.scm | 32 ++++++++++++++++++++++++++++++++
>> >  1 file changed, 32 insertions(+)
>> 
>> Oops, forgot copyright line. New patch below. Perhaps pdf.scm is better?
>
> Looks good. I think pdf.scm is better. Can you send another patch?

New patch attached.

>
>> +     ;; Prevent creation of the egg. Without this flag, various artifacts
>> +     ;; from the build inputs end up in the final python3 output. It also
>> +     ;; works around https://debbugs.gnu.org/cgi/bugreport.cgi?bug=20765 .
>> +     `(#:configure-flags '("--single-version-externally-managed" "--root=/")))
>
> Feel free to add something to that discussion. We are starting a new
> "core-updates" cycle, so it's a good time to begin implementing changes
> to core components such as build systems.

There are quite a few packages suffering from this, at least
python-{pillow,dendropy}. What's strange is that site.py and all
propagated /bin files end up in the python3 output, causing a mountain
of collisions. I'll have a shot at figuring out what's going on.

Thanks,
Marius