[opt](pointer cast) better typeid_cast #32614
Conversation
|
Thank you for your contribution to Apache Doris. Since 2024-03-18, the Document has been moved to doris-website. |
|
run buildall |
|
clang-tidy review says "All clean, LGTM! 👍" |
|
TeamCity be ut coverage result: |
|
PR approved by anyone and no changes requested. |
| #else | ||
| if (typeid(*from) == typeid(std::remove_pointer_t<To>)) { | ||
| requires std::is_pointer_v<To> | ||
| To typeid_cast(From* from) noexcept { |
There was a problem hiding this comment.
In debug mode, the previous code will print exception. It is useful to help us debug.
There was a problem hiding this comment.
I dont know whether typeid_cast(nullptr) is designed as a common use scenario
| To typeid_cast(From* from) noexcept { | ||
| if ((typeid(From) == typeid(std::remove_pointer_t<To>)) || | ||
| (from && typeid(*from) == typeid(std::remove_pointer_t<To>))) | ||
| return static_cast<To>(from); |
There was a problem hiding this comment.
maybe we could always throw exception, not return nullptr.
There was a problem hiding this comment.
maybe we could always throw exception, not return nullptr.
doris/be/src/vec/columns/column_array.cpp
Line 513 in 2c01739
the users of typeid_cast sometimes relay on the nullptr as their control flow, exception has bad performance when it is used as control flow.
There was a problem hiding this comment.
不是的，我们抛异常是在入参为空指针的时候才会抛。以前的行为，如果typeid不同，就是返回空指针。因为现在这个新写法应该不会带来更多的风险，应该是靠谱的。空指针的话肯定是代码有错才会出，这种情况下前、后的使用肯定会出问题，不必在这里处理成异常的。
zhiqiang-hhhh
force-pushed
the
opt-typeid-cast
branch
from
bec8d22 to
eaa05d2
Compare
|
clang-tidy review says "All clean, LGTM! 👍" |
4 participants
No description provided.