[reportlab-users] Installing ReportLab on Mac
Reed.L.OBrien at nasa.gov
Fri Jan 2 10:16:24 EST 2009
On Jan 2, 2009, at 9:41 AM, Andy Robinson wrote:
> 2009/1/2 Dirk Holtwick <dirk.holtwick at gmail.com>:
>> Just to drop another name into the round I would like to suggest to
>> have a
>> look on "pip" ;) Maybe this could at least be a solution for some
> This actually looks to me like a big step in the right direction. I
> wish setuptools had worked this way from day one. And if it got
> popular we could add a 'setup_pip.py'. But we won't worry about it
> for our 2.3 release. I'm taking the view that having a distutils
> "setup.py" is the python standard until Guido says otherwise; but we
> will happily include alternatives contributed by others for them to
> use at their own risk.
> IMHO "Real Men" unzip the package, put it on their path, and know what
> extensions they are compiling. Unix should seek to educate the
> developer. Sadly there are very few left.
Some of my, um, apps/systems, have dozens of interdependent packages
as well as several dozen python packages. It would be near if not
impossible to have half a dozen people unzip configure and compile all
of these things manually.
Additionally there are multiple target installation environments: OSX,
RHEL, Fedora (and sometimes ubuntu), Solaris and sometimes FreeBSD. I
can't imagine the level of effort required to maintain debs, rpms,
ports, fink, solaris whatevers and so forth.
zc,buildout (which has its own warts) with setuptools and a privately
maintained index allows maintaining a system that builds development
and production configurations, repeatably and reliably, for all of the
above platforms with minimal headache. I am not arguing others should
do it this way, but I thought I would share why I use non "Real Men"
> - Andy
> reportlab-users mailing list
> reportlab-users at reportlab.com
More information about the reportlab-users