intl extension couple of micro optimisations for error edge cases. #10044
Conversation
making c++ compile time few enums ranges.
There was a problem hiding this comment.
Thanks for the PR!
Frankly, I don't know whether marking these conditions as UNEXPECTED buys us much (PGO is not unlikely to provide even better results), but I don't think they hurt in any way.
I'm less sure about using constexpr, though. Maybe @sgolemon can comment on that.
| @@ -37,7 +37,7 @@ using icu::GregorianCalendar; | |||
| using icu::StringPiece; | |||
| using icu::SimpleDateFormat; | |||
|
|
|||
| static const DateFormat::EStyle valid_styles[] = { | |||
| static constexpr DateFormat::EStyle valid_styles[] = { | |||
There was a problem hiding this comment.
constexpr is a C++11 feature; I'm not sure whether we can require it.
There was a problem hiding this comment.
Isn't the issue that REHL 7 is still not EOL and the GCC version which comes bundled doesn't support C11/C++11 fully?
There was a problem hiding this comment.
g++ 4.8 supports the vast majority of C++ 11 features, we already use smart pointers, another C++11 nicety, since 8.2.
There was a problem hiding this comment.
Huh, well then I suppose it should be fine?
|
Well, the official Windows builds do some very basic PGO, and that could be extended (it was prior to PHP 7.4), but I doubt that a general PGO is really that useful. However, well PGO optimized applications may benefit greatly. |
making c++ compile time few enums ranges.