[...]
These are more or less real examples, I just simplified the
terms. Here are some from MZLIB:
>
return (status == MZ_OK) ? MZ_BUF_ERROR : status;
>
return (pL == pE) ? (l_len < r_len) : (l < r);
>
sym = (match_dist < 512) ? s0 : s1;
>
return ((pState->m_last_status == TINFL_STATUS_DONE) &&
(!pState->m_dict_avail)) ? MZ_STREAM_END : MZ_OK;
>
I believe that in the first three, all parentheses are superflous, but
they are used anyway. Why is that?
Obviously it's because the author of the code thought it was
clearer with the parentheses (or was working under a coding standard
written by someone who thought so). I don't think there are any
deeper conclusions to be drawn. I would have written most of them
differently, but it's not a big deal.
(My preferences for ?: are that the whole thing is syntax, outside of
the precedence scheme, and that it has mandatory parentheses. That
second line would then look like this:
>
return (pL == pE ? l_len < r_len : l < r);
>
There are fewer parentheses in all, and less potential confusion. You
can even have assignments in each branch; they will not interfere with
?:.)
But the precedence scheme *is* syntax. If you prefer to think of ?:
as something other than an operator, something that that doesn't
follow the same set of rules as other operators, and if that works
for you, then that's fine. But then how do you know that
return (pL == pE ? l_len < r_len : l < r);
means
return ((pL == pE) ? (l_len < r_len) : l < r);
and not
return (pL == (pE ? l_len < r_len : l < r));
?
I know that because I know that ?: is an operator that binds more
loosely than "==".
In any case, however you think about ?:, it's clear that
"pL == pE ? l_len < r_len : l < r" is an expression, and "return"
*is* outside of the precedence scheme. The outer parentheses are
superfluous but harmless. (Personally, I dislike parenthesizing
the expression in a return statement.)
[...]
--
Keith Thompson (The_Other_Keith) Keith.S.Thompson+u@gmail.com
void Void(void) { Void(); } /* The recursive call of the void */
Les messages affichés proviennent d'usenet.