[istream.sentry] Minor tweaks

[cpplearner]

1. Change "noskipws is zero" to "noskipws is false". The latter is more conventional since noskipws is a bool.
2. Rename the variable ctype to ct. This prevents it from shadowing the class template std::ctype.
3. Remove redundant comparison between the result of ct.is(…) (which has type bool) and 0.
4. Add a semicolon to make the sentence structure more clear.

[eisenwave]

Suggested change

	\tcode{\exposid{ok_} != false};
	\tcode{\exposid{ok_}} is \tcode{false};

The == false below should also be changed for local consistency; then we have local consistency about using "is true"/"is false" in this \itemdescr. Not fixing this would be silly if we're already touching this line.

[tkoeppe]

The original wording is awfully terse. Can we spell this out a bit more, like:

"""
If, after any preparation is completed, is.good() is true, then ok_ is true, otherwise ok_ is false.
"""

Or could we deunpack this even further and say something like "the value of ok_ is is.good()."?

@jwakely?

[jwakely]

Well we definitely can't change "ok_ != false" to "ok_ is false"!

I agree with tkoeppe's suggestion, including adding "then" after the third comma.

We could also just say:

After any preparation is completed, ok_ is set to the value of is.good().

Just saying ok_ = is.good() seems much simpler than describing the following logic, which would never pass code review:

if (is.good() == true)
  ok_ = !false;
else
  ok_ = false;

[jwakely]

Anything would be better than what's in the draft today ;)

[timsong-cpp]

Also, this sentence should probably be part of the effects instead of remarks?

[timsong-cpp]

While we are here, we can also drop the quote of the constructor signature in the previous paragraph.

Like...it's a Remarks for this constructor. Do we really need to say yet again that we are talking about this constructor?

[jwakely]

I think these are good improvements so far.

But these two paragraphs are saying the same thing, so I think it would be even more clear (and shorter) if we combined them:

Suggested change

	The currently imbued locale in \tcode{is} is used
	to determine whether or not the next input character is whitespace.
	The character \tcode{c} is a whitespace character if
	\tcode{ct.is(ct.space, c)} is \tcode{true},
	where \tcode{ct} is \tcode{use_facet<ctype<charT>>(is.getloc())}.

[tkoeppe]

Thanks @cpplearner and @jwakely for working on this -- this wording can definitely do with improvements, this is great!

[jwakely]

As this describes the final step, I think it would make more sense to be right at the end, after the "During preparation" sentence (maybe as a separate paragraph).

That way the structure would be:

• The function extracts and discards each whitespace character.
• If extracting a character returns eof, set failbit|eofbit.
• Whitespace is determined using ctype.
• The constructor may set failbit and throw. (for ... reasons? I guess the footnote covers that)
• After any preparation is completed, set ok_.

What do you think?

[cpplearner]

IMO this makes sense if we drop this \remarks macro.

draft/source/iostreams.tex

Line 4591 in 61aaf2e

\remarks

[jwakely]

Ah yes, agreed.

[jwakely]

I think this is a big improvement, thanks

[tkoeppe]

"Whether" already means "whether or not", so can we drop the "or not"?

On the other hand, "if" does not already mean "only if", so can we say "if and only if" below?

And maybe we can also change the full stop after "is whitespace" to a colon to signal that what follows is the explanation of what whitespace is?

[tkoeppe]

This is great, thanks!