[Shootout-list] Selection of Languages

[Isaac Gouy]

--- Brent Fulgham <bfulg at pacbell.net> wrote:

> --- Isaac Gouy <igouy2 at yahoo.com> wrote:
> > --- Einar Karttunen <ekarttun at cs.helsinki.fi> wrote:
> > > * How common the language is?
> > > This is fairly trivial.
> > 
> > And is "how common" a positive or negative? :-)
> > 
> > iirc Brent's intention in reviving the Shootout was
> > to draw attention to less well known languages (aka 
> > Scheme). I share that motivation, and that's why
> I've
> > worked on Clean and Smalltalk and Oberon programs. 
> 
> Part of the problem is that I'm not really clear of
> the goals myself.  I initially revived the shootout
> so I could see how the Schemes and Haskells of the
> world were stacking up against the old mainstays C
> and C++ in terms of performance.  Consequently, I
> loaded it up with lots of Schemes, Lisps, MLs, etc.,
> and only had GCC and GCJ.
> 
> Eventually, I was convinced to include commercial
> compilers (such as the Intel C/C++ and Fortran
> compilers, and the SUN Java implementation), which I
> think are useful for comparison.
> 
> At root, I think the shootout is really useful for
> quickly seeing how a given language compares with
> more mainstream options.  Consequently, it is a bit
> heavily weighted towards the more obscure languages
> (since I don't want to write C++ at home after doing
> it at work all day!)
> 
> > You'll find several Feature Requests for languages,
> > with assurances that programs will be written and 
> > sent real soon now... 
> > 
> > (Apologies to the Groovy guys who actually have sent
> > a good number of programs).
> 
> And apologies to everyone from me for not being able
> to run the shootout for the last few weeks.  My
> construction should be complete by September 1, and
> the new wiring should be done in the next few days,
> so I should be able to get a run in sometime in the
> next week or so.  (Of course, Isaac has heard this
> before ;-)

I believe it everytime :-)

(It's important to know that the intention is to restart Shootout in
the near future.)


> > > Same for four Scheme implementations. It would be
> > > nice if the language folks could pick up just a
> > > few implementations to the main page...
> > 
> > We already did that, and there were complaints.
> > We may do that again, and I'm sure there'll be more
> > complaints ;-) 
> 
> Yes.  Every time we take something off the main page
> (or try to pare things down to just a few
> representative
> copies of a language, such as Java) I get empassioned
> pleas for the language to be added back (or for one
> implementation versus another to be used as the
> representative).  It's a no-win situation.

If we only want "representative implementations" of more mainstream
languages, then we could make an "unfair" decision to move GIJ and
Kaffe and SableVM to The Sandbox (or, better still, simply remove them
completely). 

One day maybe there'll be Harmony
http://mail-archives.apache.org/mod_mbox/incubator-general/200505.mbox/%3CCA4BEB82-3D84-457D-9531-1477DD749919@apache.org%3E

Of course, the "representative implementations" of Java should include
Sun Java 5.

Similarly, maybe a Scheme maven could recommend Scheme A B & C as the
Scheme's most likely to win the hearts of newbies.


> > > Personally I would rather see Scala, Io and maybe
> > > even Felix than yet another Java implementation 
> > > added to the main page.
> 
> I agree.  Isaac has argued strongly (and convincingly)
> that we need to limit the main page to reduce clutter
> and make it possible for people to find what they
> need.  Obviously, this is not ideal since it means
> limiting the number shown on the main page.

"We" (the team) are a more basic resource constraint.

imo We should be realistic about how many different language 
implementations we can adequately support. There are always
configuration issues, there are always upgrades, there's always
puzzling behaviour that must be investigated and resolved.

Read these as examples not criticisms:
- We don't help the Pike community unless we install to do double math
- We don't help the Dylan community unless we upgrade to an
implementation that supports formated printing

These are our real limits.


> But I'm afraid I don't know how to resolve this to
> everyone's satisfaction.  No matter what we do,
> someone will be unhappy.


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com