[squeak-dev] what is holding back Smalltalk?

Keith Hodges keith_hodges at yahoo.co.uk
Fri Nov 21 04:21:14 UTC 2008


 
> In his words, he is "extremely prejudiced" against Smalltalk, even
> though he used it during a big project "back in the day."  He calls it
> a "quirky, unusual" language that you can do "great and powerful
> things" with.  The "Father of real object oriented programming," but
> only a "niche" language that small groups of bright people use to go
> of and accomplish amazing things with.  But, because the business is
> full of "ordinary" people, he thinks you would never be able to find
> anyone to work with the code you have written as a business when your
> Smalltalk developers leave for other pursuits.
I once wrote a simulator for telecoms equipment. The original demo took
2 weeks to produce in order to convince my boss.

After 3 months work, the simulator was simulating a single piece of
equipment 2 months before real equipment was available. This gave the
whole team a considerable head start. After a further 6-8 months the
simulation was doing 1000 pieces of equipment simultaneously, of 3
different varieties, while at the same time simulating up to 20 users
prodding the management system. There were 1500 unit tests ensuring that
everything was according to spec. The simulation turned out to be key to
proof of concept for our clients signing on the dotted line.

On the other side of the office, a contractor attempted to write a
similar simulator for another piece of equipment, in perl. After a year
that was scapped and a team of 4 started in Java. That was also scapped
and a top guru tried again in java, his efforts ran on 10 pcs! Finally,
last I heard another extremely expensive contractor was starting again
in C++.

I estimate (being generous) that they must have spent over a half a
million pounds on that failed project, and that doesnt include some
rather expensive bought in libraries (for which source code was not
visible). Little ol- me knocked up my Smalltalk equivalent for perhaps
5-10% of the cost. (we did buy an ST/X licence for £2000).

When I went on holiday, the only non-programmer in my team, the guy
doing automated testing was quite able to fix bugs, run unit tests and
keep things going. When I left the company the entire system was handed
over to a perl programmer, and last I heard it was still being used by 9
people daily. I must ring up and ask if it reached its 10th bithday.

So... if your boss is happy to spend 10x as much to get a poorer
result... there is not much else can be said.

cheers

Keith



More information about the Squeak-dev mailing list