Magma

Avi Bryant avi at beta4.com
Fri Jul 9 20:18:37 UTC 2004


On Jul 9, 2004, at 11:55 AM, Chris Muller wrote:

> I can think of no reason anyone should need to get into the code for 
> Magma
> anyway.  If you find a problem and you don't want to try to debug it 
> yourself,
> let me know and I'll help you the best I can.

Chris, there is exactly one reason why I would prefer to be using a 
pure-Smalltalk database like Magma, rather than a C++ server like 
GOODS: the expectation that I would personally be able to dive in and 
fix problems, or add features, if I needed to.  In either case, I could 
email the author if I wanted something - and both authors have shown 
themselves to be quite responsive.  But that's not the point.  I want 
the code to be such that I can work with it myself, without depending 
on someone else's availability and inclination.  I can't speak for 
anyone else, but if you're trying to sell me, personally, on a 
Smalltalk OODB, then you're trying to sell me on being able to hack it. 
  So, yes, I absolutely *do* need to be able to "get into the code for 
Magma", or at least to know that I reasonably could - that's the whole 
idea.

Which isn't to say that Magma *isn't* hackable - I haven't seriously 
tried.  But, I repeat, the sheer volume of code (number of packages 
(14), number of categories (52), number of classes (~200)) is an "image 
problem" in this respect.  Perhaps some overview documents laying out 
how the codebase is structured would help?  Are there any out there 
already that I've missed?

> Let me say, no matter what anyone else thinks about it, *I* will 
> continue to
> use Magma for *my* projects.  I know how reliable it is; and I'm 
> starting to
> build some serious programs on it that I will be releasing later this 
> year.

Great!  Chris, I clearly came off sounding very harsh, and I honestly 
didn't mean to.  I'm simply trying to understand, and to help you 
understand, why Magma doesn't seem to be getting the attention and use 
that it deserves to, given how much work you put into it.  If this 
discussion helps get it some more attention, and helps get rid of some 
of the lingering issues that are stopping people from using it (and 
yes, these do exist, see my "Magma notes"), then I'll have accomplished 
what I was trying to do, however caustically.

Cheers,
Avi




More information about the Squeak-dev mailing list