[Seaside] backwards compatibility vs. progress

Tim Rowledge seaside@lists.squeakfoundation.org
Wed, 26 Jun 2002 13:04:55 -0700

I'd say that you can make major changes without upsetting people by
a) delivering a reasonably clean last release of the previous design
b) carefully explaining the changes needed and advantages in updating
(or you risk leaving people behind and losing any point in doing all
that work)
c) whenever possible finding simple recipes for upgrading old code.

All of which is a bunch of work, but less pain than doing a whole bunch
of designing and writing only to find nobody ever uses it. In my not at
all humble opinion of course.

So I would suggest doing everything you can to make a really clean 0.94
available to last for a while and then put it on maintenance whilst
working on the new thing. Given the pleasantly small userbase right now,
at least you aren't going to be irritating a large number of high profit

It might also be worth spending some time on some doc and examples that
you can point people to to save having to answer so many questions. It
can also help in really working out what you want to change and why,
thus saving development time later. In fact I try to write doc first at
least to a skeleton level in order to have some idea what I want to
achieve. After all if you can't explain how to use it, what is the point
of producing it?

Tim Rowledge, tim@sumeru.stanford.edu, http://sumeru.stanford.edu/tim
Strange OpCodes: IXM: Initiate X-rated error Messages