PICAXE line limitations?

As a hobbyist myself, projects sometimes grow, an extra sensor here, more functionality there, as opposed to a more professional project with a fixed brief. Multiply that by what used to take me 600 lines I can probably do in 150 now with a couple of sub routines and a flag here and there.

How many lines? Currently I building and processing my robots maps with a dedicated LPC2104 and GNU GCC, in hind sight I should have used the LPC2106 as change in sensor now means that I could have increased the maps resolution and the code space is already at 34k. Don't get me wrong there a lot of optimisation that can be done but it's a fluid design stage not production optimisation. 600 lines for a line follower is great but how about a maze mouse or mapping a house?

I also think that although they may be saying code space what they may also be including in that statement is resources (storage and code), speed and the flexibility a native as apposed to PCODE may give. Don't get me wrong I wish the PicAxe range had been for me years ago as Intel Basic 85 on a wire wrap board was a learning curve and a half for me! They really are as cheap as chips compared to a Basic Stamp, but adding more code storage may not be the solution. Looking at the BasicX-24p for example with 8000+ lines of code user will hit a speed and flexibility issues long before storage space.

I think it's time you do what thousands of others have done, and move on from the slow and resticted pic basic interpreters.