On Mon, 12 Jul 2010 23:45:59 Bruce & Breeanna Rennie wrote:
> To all interested,
>
> The problem lies in using the conjunction & directly between the if
> expression and the following expression.
>
> If the If expression is surrounded by {} then the following
> expression is evaluated when the if does not produce &fail
>
> Hence
>
> if test then { expression1} else { &fail } & expression2
>
> doesn't evaluate the expression2 if test produces a result
>
> however
>
> {if test then { expression1} else {&fail}} & expression2
>
> does evaluate expression2 when test succeeds
>
If that's the source of your problem then it's merely a precedence
issue. if/then/else is lower than & so your 1st sample is like
if test then { expression1} else ( { &fail } & expression2 )
As a finer point, your assertion "does evaluate expression2 when test
succeeds" should actually be: "does evaluate expression2 when test AND
expression1 succeeds."
The point I'm making is, the value of if/then/else is the value of
whatever expression is evaluated - ie the { expression1 } or the { &fail
}
since { &fail } yields no value the IF yields no value and so
expression2 won't be executed.
A further simplification:
if test then expr
is identical to
if test then expr else &fail
since the missing else part implicitly means a failure.
So ................
many of us would probably write (for your sample here)
test & expression1 & expression2
Looking at your original sample, I have the following comments:
wrapping in { } is not necessary - for example
{
# this is not
write("In MatchComment()")
} &
although it is a good way to "place-hold" a block if you want to slip in
debugging statements as you've done a few lines earlier.
wrapping. Plain old parentheses are fine for this duty; unlike most
languages, even things that look like control structures are actually
just expressions, so you can wrap if, case, break, every, etc etc in
parentheses just to control the precedence.
Similarly, I don't see the point of
{
found.Pop()
}
a few lines into the code
The do-clause is optional on while, so your
do { } could be removed
Now that I see where you're coming from, the following
if found.Push(not MatchEndOfLine()) then {
{
found.Pop()
}
} else {
&fail
} &
if found.Push(MatchDOT()) then {
{
found.Pop()
}
} else {
&fail
}
would reduce (in steps) to
( if found.Push(not MatchEndOfLine()) then {
found.Pop()
} ) &
if found.Push(MatchDOT()) then {
found.Pop()
}
and finally
found.Push(not MatchEndOfLine()) &
found.Pop() &
found.Push(MatchDOT()) &
found.Pop()
So I guess the entire procedure would reduce to
procedure MatchComment()
return (
found.Push(tab(match("#"))) &
found.Pop() &
write("First Part of MatchComment()") &
write("In MatchComment()") &
while write(
found.Push(not MatchEndOfLine()) &
found.Pop() &
found.Push(MatchDOT()) &
found.Pop()
) | 1& # related to my guess below
# not sure what this is for - to yield a success
# after WHILE finally fails?
&pos &
found.Push(MatchEndOfLine()) & found.Pop()
)
end
Sooooo !
my guess at a rewrite:
procedure MatchComment()
# I like to short-circuit the flow and leave
# when it's clean like this
if not found.Push(tab(match("#"))) then
fail
# pop the matched bit to keep the stack clean
found.Pop()
write("First Part of MatchComment()")
write("In MatchComment()")
# I'm guessing a consumer loop so let it run
while write(
found.Push(not MatchEndOfLine()) &
found.Pop() &
found.Push(MatchDOT()) &
found.Pop()
)
return found.Push(MatchEndOfLine()) & found.Pop()
end
Since you seem to be discarding the matched bits because it's a comment,
I guess this specific procedure could be simplified again:
procedure MatchComment()
# I like to short-circuit the flow and leave
# when it's clean like this
if not ="#" then # =X is a synonym for tab(match(X))
fail
write("First Part of MatchComment()")
write("In MatchComment()")
while not MatchEndOfLine() do
write(MatchDOT())
return MatchEndOfLine()
end
which reads quite succinctly even with the debugging writes.
------------------------------------------------------------------------------
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
_______________________________________________
Unicon-group mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/unicon-group
