Convincing a harvester (was on SqF list)

Andreas Raab andreas.raab at gmx.de
Wed May 7 12:32:38 UTC 2003


> <whisper>3.3a modules</whisper>.

No, say it out loudly. There are lessons to be learned here but one of them
(which almost every single person here ignores) is that we had several
successes with other larger portions of work which have been done outside,
which have been done more or less by individuals (or small groups of
people). For example, who claims that he thoroughly understands how Genie
works besides Nathanael? I believe that the corpus of work he has done is
actually larger than 3.3a modules and yet it works perfectly well although
it has touched _very_ fundamental pieces of code (event handling in Morphic
for example).

The primary issue to keep in mind when dealing with such portions of work
is: How much do 'not-quite-finished' changes interact with the rest of the
system? Can you just ignore it and go on without being affected? In the case
of 3.3a modules you could not since it affected about everything in the
system. That's not true for the multi-lingualization stuff.

You should not forget that while we had a single desaster, we had numerous
successes. I actually tend to think that one of the reasons why I was all
for integration of the 3.3a modules system was how smoothly everything went
up to that point. So while being cautios is certainly a good thing, you
should also keep in mind that there are several good examples where it
worked quite nicely.

Cheers,
  - Andreas



More information about the Squeak-dev mailing list