[lib] Canonicalize order of library descriptive elements. #4067
Conversation
|
What are those 6 remaining cases? I remember @zygoloid mentioning some that needed LWG attention because their order may have been intended and reordering the elements could change the interpretation of the content being described. |
|
Remaining cases: iostreams.tex:3703-3794: bad element ordering: returns remarks effects |
Maybe I actually meant this comment by @jensmaurer: #2209 (comment). I'm still looking for it, just in case. |
|
This particular concern has been addressed by the "mandating" papers. |
|
See thread starting at https://lists.isocpp.org/lib/2020/02/15447.php. |
source/algorithms.tex
Outdated
| \pnum | ||
| \expects |
There was a problem hiding this comment.
Not really an order correction, but a merge. Shouldn't it become a list? This is how it looks like (single paragraph number, multiple grammatical paragraphs, treating the two different matters being merged):
There was a problem hiding this comment.
Similar changes in this PR didn't dropped the \expects and not the \pnum, so perhaps that would do here.
source/strings.tex
Outdated
| would exceed \tcode{max_size()} (see below), or | ||
| \item any exceptions thrown by \tcode{allocator_traits<Allocator>::allocate}. | ||
| \end{itemize} | ||
|
|
source/strings.tex
Outdated
| \begin{itemize} | ||
| \item \tcode{out_of_range} if \tcode{pos1 > size()}, | ||
| \item \tcode{length_error} if the length of the resulting string | ||
| would exceed \tcode{max_size()} (see below), or |
There was a problem hiding this comment.
max_size() is defined above, so maybe the "see below" refers to the now above Effects: for when the length would exceed max_size(), which should be redundant given [string.require]p1.
| would exceed \tcode{max_size()} (see below), or | |
| would exceed \tcode{max_size()} (see above), or |
| would exceed \tcode{max_size()} (see below), or | |
| would exceed \tcode{max_size()}, or |
source/strings.tex
Outdated
| \begin{itemize} | ||
| \item \tcode{out_of_range} if \tcode{pos1 > size()}, | ||
| \item \tcode{length_error} if the length of the resulting string | ||
| would exceed \tcode{max_size()} (see below), or |
There was a problem hiding this comment.
Same concern as https://github.com/cplusplus/draft/pull/4067/files#r449724740.
| \expects | ||
| If an implementation has strict pointer safety\iref{basic.stc.dynamic.safety} | ||
| then \tcode{ptr} is a safely-derived pointer. | ||
|
|
||
| \pnum | ||
| \expects |
There was a problem hiding this comment.
Wouldn't it be a good idea to deindent the elements a little? Although it might not be worth it considering there aren't many examples like these:
There was a problem hiding this comment.
I think we do both styles. @jwakely , any opinion here?
jensmaurer
force-pushed
the
reorder
branch
2 times, most recently
from
3b4ca85
to
32ddb62
Compare
|
@timsong-cpp , I've removed the reordering of the two mandates/constraints pairs that are the subject of the references reflector thread. |
Ah, that reminds me where it was: #3788 (comment). |
|
After the reversions, the remaining ordering violations are: iostreams.tex:3703-3794: bad element ordering: returns remarks effects |
Partially addresses #3677
(There are 6 remaining mis-orderings that need more attention.)