Empty declarations are cruft; they (may) compile, but they violate the language standards, don't contribute anything of value, and clutter up the program. Like cobwebs, they should be swept away.
=== on 24 Mar 2015, 19:03:23 Evgeny Mandrikov wrote:
\[~ann.campbell.2] we could also add that this actually violates restrictions in standard - see C11 6.7p2 and CPP11 7p3, but might be accepted by compilers.
=== on 24 Mar 2015, 19:32:50 Ann Campbell wrote:
Added [~evgeny.mandrikov]. Is there an officially hosted copy of the language standards we can link to?
=== on 24 Mar 2015, 22:07:05 Evgeny Mandrikov wrote:
\[~ann.campbell.2] ISO standards are paid.
=== on 25 Mar 2015, 13:04:35 Ann Campbell wrote:
One more question [~evgeny.mandrikov] is C11 the full, official name of the standard? Should I throw "ISO" in there?
=== on 25 Mar 2015, 18:08:20 Evgeny Mandrikov wrote:
\[~ann.campbell.2] according to \http://en.wikipedia.org/wiki/C%2B%2B - "{cpp}11" is informal name for "ISO/IEC 14882:2011", "CPP11" - is a name, which we use in code to refer to it. Similarily for "C11" - \http://en.wikipedia.org/wiki/C11_(C_standard_revision).
=== on 25 Mar 2015, 20:03:24 Ann Campbell wrote:
Thanks [~evgeny.mandrikov]
=== on 25 Mar 2015, 20:14:39 Evgeny Mandrikov wrote:
\[~ann.campbell.2] however I'm not sure that usual developers know that "9899" refers to "C" and "14882" to "{cpp}" :) but that's probably not a problem, so overall LGTM.
=== on 23 Nov 2015, 21:40:30 Ann Campbell wrote:
\[~evgeny.mandrikov] I've just mapped this to PC-Lint rule 814
____
useless declaration -- A tagless struct was declared without a declarator. For example:
----
struct { int n; };
----
Such a declaration cannot very well be used.
____
Is that case already covered, or is a ticket needed?
=== on 24 Nov 2015, 15:03:51 Evgeny Mandrikov wrote:
\[~ann.campbell.2] I suppose that this case was not covered, but you can check by yourself ;)
Also IMO such change requires more investigations, because:
{noformat}
struct {
int i;
};
{noformat}
gives warning ("declaration does not declare anything") with Clang in C mode and error ("anonymous structs and classes must be class members") in {cpp} mode, while no messages for
{noformat}
struct S {
struct {
int i; // can be accessed as "s.i"
};
} s;
{noformat}
which is probably related to \https://msdn.microsoft.com/en-us/library/a3bbz53t.aspx and \https://gcc.gnu.org/onlinedocs/gcc/Unnamed-Fields.html.
And in any case seems that current rule description is not applicable for such case, because it doesn't violate standards.
Also could be noted that seems that PC-Lint rule 814 available only for C.
=== on 24 Nov 2015, 15:17:59 Ann Campbell wrote:
\[~evgeny.mandrikov] you do realize you didn't actually answer the question, right? ;-)
Should I unmap, or do you feel this case is worth covering?
Re only-for-C, I had planned to follow up on that issue separately. In short: yes. I'm lazy.
=== on 5 Feb 2016, 16:46:42 Evgeny Mandrikov wrote:
\[~ann.campbell.2] my answer was - given current description I don't see why PC-Lint 814 maps to this RSPEC.
=== on 5 Feb 2016, 17:02:00 Evgeny Mandrikov wrote:
\[~ann.campbell.2] and to add a little bit more oil into fire of this discussion - some other cases in CPP-1371
