Re: [q-lang-users] Q 7.1 RC3 available

[Albert G. <Dr....@t-...>]

All right, FRS seems to work again, and meanwhile I cleaned up the mess 
that was left behind after my earlier attempts. Happy downloading! :)

John Cowan wrote:
> I built what I believe to be the second rc2 (size = 3012490 bytes)
> successfully on Cygwin.

The file size seems to be ok. Here's the md5sum of the latest 
q-3.1rc3.tar.gz if you want to verify that, too:
0d9b6c300ef830e301dc3025a6e73671

> A thought:  Currently the compiler will cheerfully accept rules like
> "f (x Y) = something;" even if there exists a rule "x Y" = something;",
> though the first rule will never be executed due to Q's eager evaluation.
> This is documented in the manual.  It seems to me worthwhile to warn,
> at least, about a rule like "f (x Y)" if there are any rules defining x.

But this case naturally arises, e.g., if you have "constructor 
equations". So it's difficult to determine exactly when those warnings 
really make sense. Of course I could let the compiler spit out such a 
warning anytime it sees two rules like this, but that would be more 
annoying than useful most of the time.

So IMHO these warnings only make sense if we can decide statically 
whether the equation(s) for "x Y" always succeed, i.e., whether x is 
"completely defined" in some sense. But what does "completely defined" 
exactly mean here, given that Q has dynamic typing and allows as much ad 
hoc polymorphism as you want? And even if we leave that question aside, 
and there's only a single equation defining x, we still cannot decide 
statically whether that equation will always succeed, since an equation 
may fail at runtime even if the lhs matches and there are no qualifying 
conditions at all, due to a computed (and deferred) "catch fail" on the 
rhs of the equation (which leaves us with the halting problem; d'oh!).

> I'll send you some notes privately on English-language fixes.

Great, keep 'em rolling. :) I still have to process the previous batch, 
but I'll probably get to that later tonight, after the Windows build of 
RC3 is out...

Thanks,
Albert

-- 
Dr. Albert Gr"af
Dept. of Music-Informatics, University of Mainz, Germany
Email:  Dr....@t-..., ag...@mu...
WWW:    http://www.musikinformatik.uni-mainz.de/ag