This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Getting Apple's libstdc++ debug mode into the FSF tree


>I don't like -debug=c++, because it says nothing about the library. 
>-debug=c++lib is more precise, and not too much more typing.

-debug=libstdcxx

Please help me try to keep some semblance of consistency around here....

;)

>The patch I'm testing now allows:
>
>   -debug=<checks>
>
>where <checks> is a comma-separated list of individual checking types, 
>consisting of:
>
>	c++lib	libstdc++ debug mode
>	all		everything we know how to check (silly for now; might not be so 
>silly later)
>
>Also, -debug is a synonym for "-debug=all".

Sounds good, modulo c++lib/libstdc++ -> libstdcxx

It would be nice (but it's not required for your patch) if -debug=c
meant /usr/lib/debug on linux. I'm sure darwin has a debug libc as well.

>We could also pull the concept-checking macro here, and add:
>
>	concepts	Define _GLIBCXX_CONCEPT_CHECKS to do concept-checking
>                       in the library.
>
>Then just using "-debug" will enable concept checking (since we're 
>being strict about everything else in the library, why not?).

libstdcxx-concepts

We have to try and play nice with other top-level libs. It's a pain, but
friendly relations are important.

This is an outstanding idea! I'm actually in favor of concept checking
being on by default for the debug library. It would be nice to integrate
the concept checking into the debug library. Maybe we should ask Phil
about this, and see what he thinks: I'm not quite sure it needs a
separate flag.

-benjamin


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]