[reportlab-users] Python 3000

[Robin Becker]

William Dode wrote:

> On 05-12-2008, Paul Barrass wrote:

>> Andy Robinson wrote:

>>> <with snippage>

>>> (a) the present codebase running on 2.3 to 2.6.

>>> (b) a Python 3.0 port which aims to present the same behaviour

>>> (c) a Python 3.0 port which aims to deprecate any stuff we don't need [...] in no rush.

>>>

>>> Does anyone here think (b) is worth pursuing to production quality,

>>> given that we have limited resources, and time spent on it would cut

>>> into time spent on (c)

>> For my usage (on an existing project), which I'd think was fairly 

>> typical, I'd have thought that any of your time spent working on the (b) 

>> option at the expense of the (c) option would be wasted.  I'd have 

>> thought that, with the backwards-incompatibility of Python itself, it 

>> would be better moving from Python 2.x/RL 2.x to Python 3.x/RL 3.x than 

>> 2/2 -> 3/2 -> 3/3.

>>

>> Of course, any new projects starting using Python 3.0 would be stuck, 

>> unable to use RL3, and there existing no RL2 - so I guess it depends on 

>> how long until RL3 is production ready, and how soon the early adopters 

>> will want it.

>>

>> I haven't looked at porting my own code yet, so I'm not really in a 

>> position to comment on how much work is involved, so my above comments 

>> are based on your earlier comments in the list.

> 

> Same for me

> 

> (b) could be interresting only to help to do (c), but it doesn't seems 

> to be.

> 


I think we should aim at b first since the main problem is in the back end which 
both b and c will need. I'm fairly sure the main problems in the front ends eg 
platypus, canvas are easy. At the point where we have a mainly compatible 
working front end a decision can be made as to which features should be killed, 
added re-arranged etc etc.
-- 
Robin Becker