updates vs. images -- limiting squeak to code

Martin Wirblat sql.mawi at t-link.de
Sat Oct 15 08:46:43 UTC 2005


Jonathan Kelly wrote:
...

> It's this very impression I had of sqeak that's holding me back from 
> saying, "Yeah, I want to use smalltalk". It seems to me that part of the 
> problem is that an image doesn't contain code, it's infected by it. I 
> will admit I probably haven't got into the headspace yet, if there is a 
> headspace, and there's obviously a lot I don't know, but I'm finding 
> less and less reason to think using smalltalk for my next project (and 
> it's a large project) is a good idea, if obviously very intelligent 
> people have difficulty managing code and staying current. Maybe "staying 
> current" is just another not so important mind set, but it's the one I 
> have (and almost certainly one the client has) ... :)
> 

Staying current or managing code is not a technical problem. Our problem 
is about people, and obviously these are not so very intelligent as you 
might think. If you compare Smalltalk with other languages, the closest 
thing you could do with Squeak to mimic what other languages can only 
do, is:

- Put a clean image and your application broken up in package files 
(Monticello) onto your client's site
- Have a load script, that loads these packages into a copy of the clean 
image, which then becomes "the application"
- Update (replace) from time to time packages of your application and 
rebuild a new application image

But that is only a clumsy way of what could be done by updating directly 
your original application image with the streamer and/or Monticello 
packages. If you do it this way, data in this image is preserved, a 
property other languages of course don't have, because they don't have 
an image :)

Regards,
Martin



More information about the Squeak-dev mailing list