> On 2020-08-14, at 12:42 PM, Vanessa Freudenberg <vanessa at codefrau.net> wrote:
> 

> 
> Smalltalk used to be a system that can be fully understood by a single person - truly a personal computing system. That is no longer the case.

Yes; a long time ago when the system was smaller (ST80v2 I guess) I could basically recite the source of any method in the system from memory. But back then the image ran in 1Mb. And did very little.

> 
> All the functionality we added over the years comes at the price of complexity (not to mention speed). It makes the system hard to understand. It makes it hard to see the design principles. We have not found a way to eliminate, or at least hide, any of the complexity we introduced.

It's an ever present tension. 

Stuff in the default image tends to get cared for, because it's right there blowing up when you make a bad change. Too much stuff in the default image makes for irritating bloat (cough, EToys, cough). Have stuff outside the image default image (seaside, magma, metacello, etc) and it gets broken too easily unless enough people are regularly using it to notice issues.

Change things to improve stuff or fix problems and you will inevitably break something else. And always, things get added that are incomplete and not totally thought out and/or not fully integrated with the existing code.

Unless you can afford the time to keep track of all that is going on, every time there is a new release it will be a major effort to port forward. Skip one or two releases and you are in real trouble - which is what I suggest the root issue is here for Trygve. Trying to jump form 3.1 to 5.3 ... eek. It's almost certainly as much trouble as porting from VW8.3 to Sq5.3. However, it *can* be done, with help. 

I wish, oh how I wish, that we had the resources to do proper reviews before including new things and bug fixes.

tim
--
tim Rowledge; tim at rowledge.org; http://www.rowledge.org/tim
Oxymorons: Exact estimate