llvm dev - Dec 2016 - RFC: Constructing StringRefs at compile time

On 29 November 2016 at 17:38, Zachary Turner <zturner at google.com>
wrote:
> I see, but I looked over your proposed implementation from earlier in the
> thread, and if I'm not mistaken I see this:
That's a different suggestion.
> That said, what did you think about my other proposal of the complicated
UDL
> with macro?
>
> #define LIT(x) x_string_ref_literal
> constexpr StringRef Strings[] = {LIT("a"), LIT("b"),
LIT("c")};
Why bother with the UDL?
#define LIT(x) StringRef((x), sizeof(x)-1)

-- 
Malcolm Parsons

> On Nov 29, 2016, at 9:52 AM, Malcolm Parsons <malcolm.parsons at
gmail.com> wrote:
> 
> On 29 November 2016 at 17:38, Zachary Turner <zturner at google.com>
wrote:
>> I see, but I looked over your proposed implementation from earlier in
the
>> thread, and if I'm not mistaken I see this:
> 
> That's a different suggestion.
> 
>> That said, what did you think about my other proposal of the
complicated UDL
>> with macro?
>> 
>> #define LIT(x) x_string_ref_literal
>> constexpr StringRef Strings[] = {LIT("a"),
LIT("b"), LIT("c")};
> 
> Why bother with the UDL?
> #define LIT(x) StringRef((x), sizeof(x)-1)
> 
There is subtlety though, it changes the result for: LIT(“hello\0world”).

— 
Mehdi

char buffer[100];
And it also allows LIT(buffer) to compile, whereas the UDL doesn't.

On Tue, Nov 29, 2016 at 9:54 AM Mehdi Amini <mehdi.amini at apple.com>
wrote:
>
> > On Nov 29, 2016, at 9:52 AM, Malcolm Parsons <malcolm.parsons at
gmail.com>
> wrote:
> >
> > On 29 November 2016 at 17:38, Zachary Turner <zturner at
google.com> wrote:
> >> I see, but I looked over your proposed implementation from earlier
in
> the
> >> thread, and if I'm not mistaken I see this:
> >
> > That's a different suggestion.
> >
> >> That said, what did you think about my other proposal of the
> complicated UDL
> >> with macro?
> >>
> >> #define LIT(x) x_string_ref_literal
> >> constexpr StringRef Strings[] = {LIT("a"),
LIT("b"), LIT("c")};
> >
> > Why bother with the UDL?
> > #define LIT(x) StringRef((x), sizeof(x)-1)
> >
>
> There is subtlety though, it changes the result for: LIT(“hello\0world”).
>
> —
> Mehdi
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20161129/5a923756/attachment.html>

On 29 November 2016 at 17:52, Malcolm Parsons <malcolm.parsons at
gmail.com> wrote:
> On 29 November 2016 at 17:38, Zachary Turner <zturner at google.com>
wrote:
>> I see, but I looked over your proposed implementation from earlier in
the
>> thread, and if I'm not mistaken I see this:
>
> That's a different suggestion.
llvm_strlen could be something like this:

#if __cpp_lib_constexpr_char_traits
constexpr size_t llvm_strlen(const char *S) {
  return std::char_traits<char>::length(S);
}
#elif __has_builtin(__builtin_strlen) || defined(__GNUC__)
constexpr size_t llvm_strlen(const char *S) { return __builtin_strlen(S); }
#elif __cpp_constexpr >= 201304
constexpr size_t llvm_strlen(const char *S) {
  const char *const A = S;
  while (*S != '\0') {
    ++S;
  }
  return S - A;
}
#else
constexpr size_t llvm_strlen(const char *S, size_t L = 0) {
  return *S ? llvm_strlen(S + 1, L + 1) : L;
}
#endif

-- 
Malcolm Parsons

On Fri, Dec 2, 2016 at 6:35 AM, Malcolm Parsons via llvm-dev <
llvm-dev at lists.llvm.org> wrote:
> On 29 November 2016 at 17:52, Malcolm Parsons <malcolm.parsons at
gmail.com>
> wrote:
> > On 29 November 2016 at 17:38, Zachary Turner <zturner at
google.com> wrote:
> >> I see, but I looked over your proposed implementation from earlier
in
> the
> >> thread, and if I'm not mistaken I see this:
> >
> > That's a different suggestion.
>
> llvm_strlen could be something like this:
>
> #if __cpp_lib_constexpr_char_traits
> constexpr size_t llvm_strlen(const char *S) {
>   return std::char_traits<char>::length(S);
> }
> #elif __has_builtin(__builtin_strlen) || defined(__GNUC__)
> constexpr size_t llvm_strlen(const char *S) { return __builtin_strlen(S); }
> #elif __cpp_constexpr >= 201304
> constexpr size_t llvm_strlen(const char *S) {
>   const char *const A = S;
>   while (*S != '\0') {
>     ++S;
>   }
>   return S - A;
> }
> #else
> constexpr size_t llvm_strlen(const char *S, size_t L = 0) {
>   return *S ? llvm_strlen(S + 1, L + 1) : L;
> }
> #endif
>
...and we've gone in a loop in this conversation, haven't we? Isn't
this
again where you end up with really slow code for compilers that end up in
the last #else, such as MSVC?

+1 from me for the StringLiteral proposal from a few messages back.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20161202/248eb92b/attachment.html>