John Doty <jpd@[EMAIL PROTECTED]
> wrote:
> Bruce McFarling wrote:
> > But this is the same old rehash ... you are continuing to ignore the
> > implications of your characterization of the state of affairs that
> > are inconvenient to your conclusion that there is something
> > intrinsic to Forth that makes it impossible for it to have a
> > publication culture, so that the only recourse is to fork an
> > incompatible version of Forth that lacks that obstacle.
>
> If anyone can overcome the obstacles (plural, I think) without forking
>
> the language, I will cheer. I believe it impossible. But I didn't
> think the Rockies had any chance of getting to the Series last year
> either: sometimes it's fun to be wrong.
You can find something that a lot of current Forth users will go along
with. Or you can find something that a lot of future Forth users will
adopt. I don't care which, but the latter is probably easier. It doesn't
matter how good your Forth is if you can't get anybody to use it.
We have some examples of Forth getting an early start in an area and
getting rejected. For example, when MUDs were pretty new somebody
released a MUD in Forth, that let privileged users do Forth scripting.
They didn't like it at all and switched to something else as quickly as
they could. I'd kind of like to see what's left from the criticisms from
all that. I remember I looked at it a little bit and it didn't look like
a particularly good Forth, but these weren't Forth connoisseurs. They
were people who liked to play roleplaying games who had a chance to
learn an easy scripting language to code their own artifacts and mobes.
And they didn't like it.
Then there's OpenFirmware. Did a lot of people learn how to use
OpenFirmware? Did they learn how to debug with it? Did they reject it,
or did the money dry up and they moved on to something that paid? I
don't have the numbers. Was it a small enough niche that not many people
got exposed to Forth? Did the coding and debugging etc methods ****ne
through, or did a lot of people stumble around figuring out for
themselves how to get results? Did we wind up with people who were glad
to quit using it, or did they tend to see how it was great for that
niche but they didn't see how too apply Forth to anything else when the
niche eva****ated?
> > The state of affairs you present is:
> > * Most code in most languages is not published
> > * Developing a publication culture around a project in a given
> > language is therefore a rare event
> > * Forth is a niche language, as are most programming languages.
>
> But it wasn't such a niche language a quarter century ago. It failed
> then to create a sustainable code base, despite the fact that it was
> fairly popular in the academic/research community. Your argument is a
> little stronger now, but you have to ignore history to make it.
It's true now. We can argue about what went wrong in the past, about
which things were historical accident and which were things that we did
wrong and which were failings in the language. I remember not long after
that there were Forth programmers who said that Forth was a sort of
skill amplifier. If you were good it made you better, and if you were
mediocre it made you worse. If I was a manager who didn't have a clear
idea how to get good programmers, I think I'd rather have something that
was dependably adequate than something that increased the variability of
results.
> Yes. Species populations have a power law distribution. Star
> luminosities have a power law distribution. But for an individual
> species or star its place in the distribution is not random: it
> depends on intrinsic properties of the species or star. So, what are
> the intrinsic properties of Forth that consign it to a tiny niche and
> how do we fix them?
Forth would need a chance. A cir***stance where a lot of people would
try it out. Right now we don't have that, hardly anybody new is giving
it much thought. I have the idea that of the people who do give it a
quick look -- which includes a significant number of CS guys -- most of
them don't give it enough of a look to figure out how to actually do
stuff with it. Like the discipline to write one- and two-line routines.
XP guys do that but they don't do it in Forth, and would somebody who
happened to know about XP carry over his skills to Forth? Or would he
have to learn all over again what works? The more usual approach might
be to look it over, see some of the cute tricks, and think that it's
hard to program in it.
If the chance was available, then Forth failings would matter. Without
that chance Forth won't get enough attention for any failings to cause
trouble. Beyond the "problems" of no enforced typing, variables
discouraged, postfix, etc. All of them problems that would disappear
after a week of extensive practice, but that most people would not put
in the time to get past.
> > both
> > because of the associated network economies and because of the
> > economies of scope, so some small number of languages will be the
> > mainstream languages ... and they could well be the very best ones
> > available, or just the lucky ones, or some mix of the two by various
> > degrees.
>
> I don't think it's luck.
Whether or not it's luck, to get a second chance we'd need a specific
cir***stance that encouraged people to try it. And Forth does take a
somewhat different mindset from usual programming. How do we get people
to pick that up quickly? Elizabeth Rather's 1-week course does it, but
to get a lot of new Forth programmers we'd need a cheaper way that
reaches many more newbies.
|
