RE: [boost] Formal review of "Output Formatters" library begins today

From: Reece Dunn (msclrhd_at_[hidden])
Date: 2004-09-14 04:29:14

Paul A Bristow wrote:
>| [mailto:boost-bounces_at_[hidden]] On Behalf Of Reece Dunn
>| >   "Boost prefers clarity to curtness".
>|
>| I have been giving this some thought. What I am looking at at
>| the moment is
>| something like:
>|
>|    std::pair
>|    <
>|       std::pair< int, std::vector< char > >,
>|       std::list< boost::math::quaternion< float > >
>|    > mct;
>
>Which leaves me puzzling what on earth mct can be short for !
>myVeryContrivedContainerType?
>
>Is this a complex demo assemblage?

Yes. It was to give an example of the new (planned) naming scheme to output 
mct [my_complex_type].

>|    namespace io  = boost::io;
>|    namespace fmt = boost::io::format;
>
>using boost::io::format;
>
>would allow you to replace fmt with format?  Which is much easier to
>understand,
>especially for those for whom english is not their first language?

I would prefer to use a namespace alias where possible as I dislike 
unnecessarily polluting the global namespace. As to
   namespace format = boost::io::format;
vs
   namespace fmt = boost::io::format;
you can see that these are aliases for "boost::io::format", just like
   namespace fs = boost::filesystem;

As long as it is put at the top of the code examples/tests and put in the 
documentation introdution at the beginning, clearly stating that "fmt" is a 
namespace alias for "boost::io::format" I don't see it being a problem.

The problem with the current implementation naming scheme is that it uses 
Xfmt consistently and most importantly, *the user has no control over this*. 
Feel free in your code to use:
   namespace format = boost::io::format;
or equivalent. Is there a concensus for/against the use of the shorter 
fmt::basic(), etc. in the docs and code? Dave? John?

>Have I commented this correctly (and informatively)?

Yes. Although I am not sure of the behaviour of using on a namespace name 
(don't you need a using namespace or a namespace alias?)

>|    std::cout << io::object( mct,
>|       fmt::pair
>|       (
>|          fmt::pair( fmt::basic(), fmt::container()), // basic used for
>int, container is a vector.
>|          fmt::container( fmt::nary()) // container is list, and n in the
>nary is 4 for a quaternion.
>|       )
>|    );
>
>
>| With io::delimiter holding one of the delimiter values.
>| io::wrapped_delimiter will replace io::openclose_formatter in holding
>open/close delimiter pairs
>
>why wrappED?  io::wrapper? ({} [] ...

The io::wrapped_delimiter above is a holder for open and closing pairs, so 
io::wrapper_delimiter doesn't seem syntactically correct. Maybe 
io::wrapp*ing*_delimiters?

>| and io::sequence_delimiter replaces io::formatter.
>
>io::separator?  , | ...

Maybe io::sequence_delimiters. This is because it contains open/close *and* 
separator delimiters, since it is controlling how a sequence type is 
rendered.

>(Of course, if English spelling made any sense whatever, it would be
>separater - but it hasn't got to be the premier language by being phonetic,
>consistent and logical, has it?)

Causing me a great deal of confusion (even being a native English speaker!)

>Paul
>
>PS
>/rubbing_it_in_mode on
>
>some filenames have the spelling mistake too ;-)
>This may confuse some using search later.

Doh!

>/rubbing_it_in_mode off
>and finally
>
>   std::cout << boost::io::formatob( vec ).format( "{ ", " }" );  
><<<<<<<<<<<
>isn't this missing a ':' separator???
>    // output: { 1 : 2 : 3 }

Thorsten spotted this as well. I have updated the docs to:
    // output: { 1, 2, 3 }

Thanks,
Reece

_________________________________________________________________
Stay in touch with absent friends - get MSN Messenger 
http://www.msn.co.uk/messenger