Feature #53
[Polysemy]Interpretation goes undetected
| Status: | Closed | Start date: | ||
|---|---|---|---|---|
| Priority: | Low | Due date: | ||
| Assignee: |  Alain Prouté | % Done: | 0% | |
| Category: | Design | |||
| Target version: | - | |||
| Platform: | Triage Stage: | |||
| Resolution: | fixed |
Description
The final line:
Has two possible interpretations and yet is compiled without error or warning.
Seemingly, one possible interpretation goes undetected.
Admittedly my code is weird, certainly i have defied some implicit recommandation that the * operator must be associative, that is of the type ($A,$A) -> $A.
<pre>
define ($A,$A,$B,$B) -> ($A,$B)
($A,$A) -> $A f * ($B,$B) -> $B g
=
($A a1,$A a2,$B b1,$B b2)
|->
(f(a1,a2),g(b1,b2)).
define ($A,$A,$B,$B,$C,$C) -> ($A,$B,$C)
($A,$A) -> $A f * ($B,$B,$C,$C) -> ($B,$C) g
=
($A a1,$A a2,$B b1,$B b2,$C c1,$C c2)
|->
with a = f(a1,a2),
with bc = g(b1,b2,c1,c2),
if bc is (b,c) then
(a,b,c).
define ($A,$A,$B,$B,$C,$C) -> ($A,$B,$C)
($A,$A,$B,$B) -> ($A,$B) f * ($C,$C) -> $C g
=
($A a1,$A a2,$B b1,$B b2,$C c1,$C c2)
|->
with ab = f(a1,a2,b1,b2),
with c = g(c1,c2),
if ab is (a,b) then
(a,b,c).
type Interval($T) :
empty,
between($T lower,$T upper).
define (Interval($T),Interval($T)) -> Interval($T)
intersect(($T,$T) -> Bool less)
=
(Interval($T) a,Interval($T) b)
|->
if a is between(a1,a2) then
if b is between(b1,b2) then
if less(a2,b1) then empty
else if less(b2,a1) then empty
else if less(a1,b1) then between(b1,a2)
else between(a1,b2)
else empty
else empty.
define
(
Interval($A),Interval($A),
Interval($B),Interval($B),
Interval($C),Interval($C)
) ->
(Interval($A),Interval($B),Interval($C))
intersect_triple
(
($A,$A) -> Bool a_less,
($B,$B) -> Bool b_less,
($C,$C) -> Bool c_less
)
=
intersect(a_less) * intersect(b_less) * intersect(c_less).
</pre>
History
#1
Updated by Alain Prouté over 16 years ago
- Status changed from New to Closed
- Resolution set to fixed
Thank you very much Damien for this remark.
However, your text is not ambiguous, but this is for a purely syntactical reason. Indeed, the binary operation * associates to the right, so that
intersect(a_less) * intersect(b_less) * intersect(c_less)is read as:
intersect(a_less) * (intersect(b_less) * intersect(c_less))making the third paragraph:
define ($A,$A,$B,$B,$C,$C) -> ($A,$B,$C)
($A,$A,$B,$B) -> ($A,$B) f * ($C,$C) -> $C g
= ...
useless, as you can check by inhibiting it. If you put a pair of parentheses like this:
(intersect(a_less) * intersect(b_less)) * intersect(c_less)the second paragraph becomes useless instead of the third one.
A similar situation is the term:
1 + 2 * 3which would have two interpretations (1 + 2) * 3 and 1 + (2 * 3) if no precedence/association rule was in effect. (In this case, this is a precedence rule, and an association rule in your example).
Behind this fact, there is undoubtly a problem, which is that precedence/association rules are rules which are in some sens implicit. If the grammar was written without theses rules, we would have to write either (1 + 2) * 3 or 1 + (2 * 3) and 1 + 2 * 3 would become a syntax error.
Recently, I wondered if it was a good idea to remove all precedence/association rules. I will not do it for Anubis 1 anyway, because it would generate big problems in the very big already exiting programs. But for Saunders, I will reduce these rules to a bare minimum. Probably, I will keep 'obvious' rules, like the precedence of * over +, but in any case, the number of rules will be small so that it is easily remembered by users.
A long time ago, I had a bug in a C program, which took me 3 days to discover. I had written something like this (I don't remember exactly):
a & b >> c
According to the rules of ANSI C this is interpreted as: a & (b >> c), but I was thinking to (a & b) >> c. I'm almost sure that nobody knows the big table of precedences/associations for C by heart.
This is a danger, and I want to reduce this danger in the future at least for Saunders.