[language.support] review library indexing of clause 18 #840
Conversation
This change consistently applies to new indexlibrarymember macro in places missed by the first pass - mostly for cases where either the class or member-name has an underscore. It consistently uses a % to eliminate potential whitespace being introduced by an index entry macro. It removes a duplicate entry for 'numeric_limits' causes by the use of the \_ command. This appears to be the correct fix, on inspection of the index. The alternative would have been consistent use of \_ on every numeric_limits entry, so trying the simpler fix first.
AlisdairM
force-pushed
the
review_clause_18_index
branch
from
f64d0cb
to
3bb6f12
Compare
|
Hang on, there are more of these issues. The problem was that I only searched for things ending in a |
|
OK. Just so you know what I am looking at: I plan to perform a clause-by-clause review, as time permits, checking every itemdecl has an index entry, and all index entries have a trailing %. For good measure, I will promote existing members to indexlibrarymember if they are not already (i.e., were not duplicated to begin with). If I find more entries introduced by codeblock rather than itemdecl, I will fix those too - but I don't have an easy way to audit for those - that is more spot it by luck as I go. |
|
Rebase please! :-) |
|
OK - it looks like the overwhelming majority of uses of indexlibrarymember are: While the latter is my preference, I will apply the former consistently (distinctly fewer changes), and make that part of my clause-by-clause audit. |
|
It doesn't matter at all; either version works. I just went with the search-and-replace that was easy (picking the first line). If you're enjoying this sort of no-op work, feel free to switch the arguments around :-) |
|
I would rather have a consistency (it aids later tooling - if any) and want to minimize churn on make-work tasks - some folks may be sensitive to the doc history, It seems putting a class second (unlike when handling constructor/destructor) is the default, based on the pre-existing order when you ran your regex. |
|
Well, I think the only thing that's consistent so far is that nothing is consistent :-) The previous commit, which used the same regex, had it the other way round. I really don't care whether a) we do anything at all or b) which way round you make it; do whatever you think is useful. |
|
Are you reusing this PR? |
|
Please reopen when it's ready. |
This change consistently applies to new indexlibrarymember
macro in places missed by the first pass - mostly for cases
where either the class or member-name has an underscore.
It consistently uses the notation of {class}{member} when
using the new macro. This is purely a consistency concern,
as the macro will expand to the same results either way.
It consistently uses a % to eliminate potential whitespace
being introduced by an index entry macro.
It removes a duplicate entry for 'numeric_limits' causes
by the use of the _ command. This appears to be the
correct fix, on inspection of the index. The alternative
would have been consistent use of _ on every numeric_limits
entry, so trying the simpler fix first.