[lib] Remove 'Constructs an object of type ...' phrases #3507
Conversation
source/numerics.tex
Outdated
| @@ -4241,8 +4228,7 @@ | |||
| \begin{itemdescr} | |||
| \pnum | |||
| \effects | |||
| Constructs a \tcode{seed_seq} object | |||
| as if by default-constructing its member \tcode{v}. | |||
| Default-constructs its member \tcode{v}. | |||
There was a problem hiding this comment.
"Default-initializes the member \tcode{v}." would be better here, but we should instead noexcept = default the declaration in the class synopsis and remove the detailed specification completely:
- defaulted constructors default-initialize non-static data members with no default member initializers, so we needn't spell out that
vis default-initialized vector's default constructor is conditionallynoexcept(noexcept(Allocator())), which istrueforstd::allocatorand therefore forv's typevector<result_type>, so this constructor should benoexceptrather than "Throws: Nothing." per LWG policy. (I assume no one has noticed that we could strengthennoexcepthere since the conditionalnoexceptwas added tovector's default constructor.)
There was a problem hiding this comment.
Please open a separate editorial issue for that. Let's not accumulate too much burden on this one.
source/time.tex
Outdated
| initializing \tcode{zone_} with \tcode{traits::default_zone()} and | ||
| default constructing \tcode{tp_}. | ||
| Initializes \tcode{zone_} with \tcode{traits::default_zone()} and | ||
| default constructs \tcode{tp_}. |
There was a problem hiding this comment.
default-initializes? (There are several occurrences of "default constructs" herein that I would prefer to be "default-initializes", I won't mark the remainder under the expectation that we'll consistently change all or none of them.)
There was a problem hiding this comment.
Make it a separate editorial issue, please. I'm trying hard to limit this pull request to the "blatantly obvious" changes, ignoring all the other wording nightmares in this section.
| @@ -5896,7 +5892,7 @@ | |||
| \begin{itemdescr} | |||
| \pnum | |||
| \effects | |||
| Constructs an object of class \tcode{bitset<N>} as if by: | |||
| As if by: | |||
There was a problem hiding this comment.
LWG has been replacing "as if by:" with "Equivalent to:" in newer wording, but for cases like this (constructor is equivalent to other constructor) we simply use constructor delegation in the class synopsis and avoid prose specification.
There was a problem hiding this comment.
-> separate editorial issue, please.
There was a problem hiding this comment.
We have historically considered changing "As if by" to "Equivalent to" to not be editorial, because "Equivalent to" has implications for the other descriptive elements as described in [structure.specifications]p4 which are not (clearly) implied by "As if by" wording.
jensmaurer
force-pushed
the
c29
branch
2 times, most recently
from
c8cacca
to
21f1037
Compare
jensmaurer
added
the
needs rebase
for constructors; this effect is implied by the core language. Only simple phrases are removed; more complex sentence structures are left unchanged.
|
Rebased. |
jensmaurer
removed
the
needs rebase
for constructors; this effect is implied by the
core language. Only simple phrases are removed;
more complex sentence structures are left unchanged.
Partially addresses #3506.