[reportlab-users] Status of Python 3 port?

Robin Becker robin at reportlab.com
Thu Feb 4 12:58:26 EST 2010


On 04/02/2010 17:34, Peter wrote:

> On Thu, Feb 4, 2010 at 5:29 PM, Andy Robinson<andy at reportlab.com> wrote:

>>

>> What I really wanted to do is make a clean break and base a new,

>> incompatible, smaller and cleaner 'reportlab3' on Python 3.x. The

>> clean distinction between Unicode and byte arrays would be a big plus

>> for our kind of library, and an incompatible new version would let us

>> do some much better things. But Guido said "don't do that" :-(

>

> How about you make a new clean lean reportlab which works on

> both Python 2.x (say Python 2.5+) and Python 3.x?

>

> Peter

........

I think python 2.x and python 3.x are fundamentally incompatible when it comes
to serious string/byte data formatting.

Even if we assume that everywhere we've done string manipulation in the current
were turned into unicode we still have lots of places in the output phase where
we'd have if isinstance(x,unicode). unfortunately that is invalid in 3.x.
Probably that indicates a failure in our back end model of pdf ie some of the
objects are a bit amorphous and generic, but then PDF has extremely generic data
structures.
--
Robin Becker


More information about the reportlab-users mailing list