[reportlab-users] ReportLab 2.2 deployment options

[Stephan Richter]

On Tuesday 02 September 2008, Andy Robinson wrote:

> 2008/9/2 Stephan Richter <srichter at cosmos.phy.tufts.edu>:

> > On Monday 01 September 2008, Andy Robinson wrote:

> >> Q1. How badly do people want setuptools, eggs, easy_install and so on?

> >> Is plain old distutils good enough?

> >

> > Very bad. In our big Zope 3 packages test suite, I have to deactivate

> > z3c.rml tests, because of a lack of a proper ReportLab package.

>

> Is setuptools now a "de facto standard" in the Zope 3 world?


Yeah. We use it and eggs on PyPI exclusively. This is probably the reson the 
Zope project produces about 25% of all packages in PyPI. (Though for 
backwards-compatibility we still produce a large tar ball.)

See here for some example:
http://svn.zope.org/z3c.rml/trunk/setup.py?view=auto

There are also several other examples in the Zope repository that compile 
C-extensions.


> >> Q2. Should we try to do "batteries included", with the accelerator,

> >> optional C libraries for rendering bitmaps and the type 1 fonts

> >> included in the basic package and one setup script to build it all?

> >

> > setuptools supports extra-includes, where you can specify additional

> > dependencies on extended functionality.

>

> OK, I see that, but it's not the important question.  What matters most

> now is how many people actually want a package without the C extensions.

> I suspect this will be a small number (doing Jython, AppEngine,

> Obscurix-1982-with-dodgy-C-compilers etc) and

> we can probably handle them with a short paragraph on the web site

> and readme saying "it's OK to delete the following...." - or a separate

> "python-only" distro if enough ask for it.


Oh, I see. Fools. :-) I prefer the C extensions of course.

Regards,
Stephan
-- 
Stephan Richter
Web Software Design, Development and Training
Google me. "Zope Stephan Richter"