Re: [boost] Proposal: Identifying libraries

[Beman Dawes]

At 01:10 PM 12/16/2002, Peter Dimov wrote:

>From: "Beman Dawes" <[EMAIL PROTECTED]>
>> At 11:15 AM 12/16/2002, Douglas Gregor wrote:
>> >Sounds good. I think we should try to do the conversion automatically,
>at
>>
>> >least for all files in subdirectories (e.g., the file boost/xxx/yyy.hpp
>> >would be assumed to be a part of library xxx if libs/xxx exists) and
>for
>> >files boost/xxx.hpp where there is a library in libs/xxx. We don't want
>> >to do this conversion by hand :)
>>
>> Good point! I'll whack together a little program to do the above (after
>> waiting awhile to make sure such a change is acceptable to others.)
>
>I agree with the general intent of the proposal (in fact I've been using
>direct links to the .html file with the documentation for some time now)

I've also been doing that too in my non-Boost code for several years. Originally I had an IDE plugin to invoke a web browser when clicked, but of course program editors now have that as a builtin feature. Works great.

>but how does this affect (implementation detail) headers that aren't
>necessarily part of a specific library? boost/detail/lightweight_mutex.hpp,
>boost/assert.hpp, boost/throw_exception.hpp, for example.

Good question. It seems to me every header should have a home library. Probably "utility" for your examples.

If "utility" accumulates several entries in a given category, as has already happened with iterators-related stuff, we should then move those components to their own library.

Note that while I agree with Doug that it would be easier to initially set up as many files as possible automatically, that is just a labor saving initialization. It isn't meant to override Boost developer wishes.

--Beman


_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost