* Update C.83 with swap for resource mgmt.
The swap is useful to implement assignments idiomatically (e.g. copy-swap idiom).
With the current enforcement, to non-virtual classes, very simple classes (e.g. `trivially_copyable`, a struct encapsulating an stl container) are suggested to have a swap. This suggestion can be argued wrong since for very simple classes copy-swap idiom is not efficient (the creation of a third object on assignment is not needed, which copy-swap does).
* Rework C.83 according to PR comments.
The various guidelines now have consistent Enforcements.
All C-style casts are now consistently banned, including to `(void)`.
Cast to `(void`)` should be `std::ignore =` instead.
All functional-style casts are now consistently banned, instead of
`T(x)` use `T{x}`.
In English, the word "may" is overloaded and ambiguous. This commit
changes it to "might" wherever possible, otherwise more specific
meanings like "can," "could," or especially "must" ("may not" -> "must
not" when that is intended).
The examples in SF.12 are likely to encourage readers to always use the `""` form of `'#include` when including headers from the same project ([discussion](https://github.com/isocpp/CppCoreGuidelines/pull/1596#issuecomment-673266275)). However, in larger projects this may not always be appropriate; `<>` should be used for includes located via a header search path.
This proposed solution adds an example of the later, i.e. where `<>` is used to include a header from the same project.
- Add a cross-reference to C.139 and note that it doesn't matter whether
a function is declared with override or final if the whole class is
already final.
- Fix C.139 to make it clearer that it's about `final` on classes.
explicitly note that "file" refers to as it exists in the location of it being authored/modified.
this is to address any confusion about #including "bar.h" from foo.h based on whether bar.h is relative to foo.h at the time and location foo.h is being modified (e.g. under mylib/foo.h) versus at the time/location from which foo.h is installed to the system and included by the user (e.g. from /usr/include).
Part of SL.str.1 references replacing “const string*” with “string_view”, but the example code above shows “const string&” instead.
This changes the comment to the reference rather than pointer.
As a general rule, `operator()` should be const-qualified.
The exceptions are few and far between.
I was actually looking for a Guideline on that subject;
I grepped for `operator()` and found that not only is there
no such Guideline yet, the doc actually contained these
places that (needlessly) violated the general rule.
Herb Sutter
shaneasd
Florian Behrens
peno2
Sergey Zubkov
David Leal
Niels Dekker
Fernando J. Iglesias García
Gabriel Dos Reis
David B
rjhcnf
John McFarlane
LE GARREC Vincent
cadenzasong
apenn-msft
Hugh Macdonald
al-mission-2016
Arthur O'Dwyer