Update the CTRE library #388
Conversation
| constexpr ctll::fixed_string<A + B> operator+(const ctll::fixed_string<A>& a, | ||
| const ctll::fixed_string<B>& b) { | ||
| char32_t comb[A + B + 1] = {}; | ||
| for (size_t i = 0; i < A; ++i) { // omit the trailing 0 of the first string | ||
| comb[i] = a.content[i]; | ||
| } | ||
| for (size_t i = 0; i < B; ++i) { | ||
| comb[i + A - 1] = b.content[i]; | ||
| comb[i + A] = b.content[i]; | ||
| } | ||
| // the input array must be zero-terminated | ||
| comb[A + B] = '\0'; |
There was a problem hiding this comment.
- Was the 0 part of the string length before and now it isn't?
- Do we use
++iconsistently instead ofi++? I am aware that there is a difference between the two, when the value is used. Is there a difference when it is not used?
There was a problem hiding this comment.
-
Yes exactly, in the new
ctreversion "a" becomesfixed_string<1>(previouslyfixed_string<2>). We have exhaustive tests that would horribly break if this method was wrong. -
If the value is used, then post- vs. pre-increment makes a huge difference, as you have correctly stated (has been like this "forever" in C. When the value is not used, then
++iis never less efficient thani++. For builtin types, likeintetc. it is exactly the same. But if the incremented objectiis e.g. an iterator to a complex data structure, which stores several pointers etc., theni++has to create a copy of the iterator (we need to old value to possibly still use it, and the new value, because we requested an increment. For this reason, I have started to use++ias a default in for loops (of course only where range-based for loops don't apply.
| fixed_string("[\\+\\-]?([0-9]+\\.[0-9]*") + ExponentString + "|" + | ||
| ExponentString + ")"; |
There was a problem hiding this comment.
I appreciate the explicit fixed_string instead of the abbreviation fs.
Is fs still used elsewhere?
There was a problem hiding this comment.
the fs is not used, and already is deleted in the code you are looking at.
Updated the CTRE library to the current release. The most visible effect of this is getting rid of compiler warnings.
It also includes small adaptations of the CTRE-based Turtle Parser, because the internal semantics of the
ctll::fixed_stringhave slightly changed.