[Ocaml-biz] The strategic future of OCaml for 2..4 years

[Brandon J. Van Every]

Brian Hurt wrote:
> Brandon J. Van Every wrote:
>
> > Really.  Ok, tell me what I want?  Your view of what I want might
> > clarify something here.
>
> You want people to stop spending time on 'frivolous' projects like yet
> another make replacement.

No, because I do want GNU Make replaced.  Ideally with an all-OCaml
coded solution.  The R&D on alternatives is not frivolous, it has to
continue.

> There should be one make, and everyone should use it.

Yes.  It would greatly aid OCaml's deployment and popularity to have one
high quality build tool that most people use.  I believe the Python and
Java communities have such things, for instance, but I'm not entirely
sure.  C# as well, taking the Visual Studio toolchain culture into
consideration.  I think we should examine what the competitors have
(Python, and who else?) and the sources of converts have (C++, Java,
C#).  It may give clear indication that unified build tools are the way
to advance a language.  Or, it may give evidence on what's a sane spread
of options, how the food chain ultimately evolves.

> We should have a five-year plan, with deadlines.

We should have deadlines, and tangible, executable plans.  That is the
nature of business, and I'm not on this list for hobby.  I want cash
flow for my trouble.

I think a 5 year plan is too much to ask right now.  If we succeed at a
1 month plan, and that leads to a 6 month plan, and we actually do it,
I'll say we're doing something right and getting farther than the Python
crowd did.  Execution has to happen.  It isn't good enough to leave this
to open source types to "just get there" without any central push.  I
think if you look at Python, you'll see evidence of a strong historical
push, from Guido especially.

> You want there to be central standards for everything.

For core things, definitely.  That's how you get "play easily on any
platform," one of Python's major sales points.  Python has a *huge* set
of standard libraries.  Lotsa things all the Python guys can count on
being there and working.

Beyond a core, standards are unrealistic.

Also the core has to evolve.  Note that OCaml hasn't succeeded in
resolving what its standard libraries are supposed to be, how to package
them, or what role INRIA and third party businesses should have
regarding them.  I don't think I want to face this can of worms right
now though.  I agree with Tony, we need *SHOWCASE* projects.  Tool
standards should be driven by popular, noteworthy OCaml projects, not
the other way around.

> You want Ocaml marketed to
> hard-headed business people, not just techies.

Yes.  Show Me The Money.

> In short, you want things
> to work like Microsoft or Java.

I think your statement here is too vague / debateable.  Let's keep it to
what I said above.

> But what gets us to concentrate is not flashy graphics

As a 3D graphics guy and game developer, I'd beg to differ.

> or cute logos or clever buzzphrases.

Those don't hurt.  BSD had its daemon, so Linux had to have a penguin.
You can most certainly market to techies.  They're just the kind of
market that hates being marketed to.  Or says they do.  In marketing,
the proof is always in what people *do*, not what they say they do.

> It's working, usefull code.

Too vague.  We need more perceptions about what is considered 'working'
and what is considered 'useful'.  I do mean *perception*.  You might
think it's about realities... I can assure you that Sourceforge is
driven by perception, people flogging off about all that great code out
there.  Little of it is actually useful.  Most of the projects are ego
projects, and NIH is rampant.


Cheers,                         www.indiegamedesign.com
Brandon Van Every               Seattle, WA

20% of the world is real.
80% is gobbledygook we make up inside our own heads.