Re: [boost] [quickbook] diagnosing a missing [endsect]?

Subject: Re: [boost] [quickbook] diagnosing a missing [endsect]?
From: Peter Dimov (lists_at_[hidden])
Date: 2017-03-22 03:00:28

Daniel James wrote:
> > My default toolset is msvc-8.0, and it seems that std::wcout doesn't 
> > quite work there, although the trick with _setmode(_fileno(stdout), 
> > _O_U16TEXT) does enable wprintf et al.
>
> Looking at the code I expected UTF-16 to work for Visual C++ 7.1 and up, 
> so I'm going to download Visual Studio 2008 and give it a go.

Not sure you should bother with that; I doubt that anyone uses 8.0 any 
longer (which is 2005 by the way, 2008 is 9.0). I'll bump my default to 
something modern.

I'm more interested in our original topic, that is, there to be a way to 
validate a .qbk file such that (a) the warning for a missing [endsect] to be 
an error (and the same applies to other similar warnings, if there are any) 
and (b) unmatched [section]/[endsect] in an included .qbk to also be treated 
the same way if the .qbk marked as self-contained in some unspecified way 
(not sure that a docinfo block is the best way to mark for our purposes, 
although I'm not exactly an expert.)

People would then be able to add a doc validation step to Travis, or to 
change their doc Jamfiles to enable strict mode so that "b2 doc" fails 
properly when it detects such mistakes.