2013/1/16 Zack Weinberg <za...@panix.com> > On 2013-01-16 2:38 PM, Benoit Jacob wrote: > >> For starters, the standard C++ <cmath> library is already much better. >> > > I was under the impression that the <cwhatever> headers could not be used > in Mozilla code. Is that no longer the case? >
Really? I've been using them in all my Mozilla code for 3 years now and never had a reviewer complain about that. What would be the rationale for allowing foo.h and disallowing cfoo anyway? > > I also expect that <cmath> should contain at least as much weird shit as > the corresponding <math.h>, since as far as I know no OS vendor has ever > actually succeeded in making the <cwhatever> headers define only symbols in > namespace std. Thus it would not help with my particular problem. However, > the math headers are a bit of a special snowflake and I could *imagine* > <cmath> not being as troublesome as <math.h>. Does anyone know for sure? I was only addressing the "surprising behavior when everything isn't a 'double'" part of your email. Clearly, <cmath> typically #includes math.h so you get the same things from it, and more. At least here on GCC 4.6 / linux, <cmath> #includes math.h. That said, I have been including it in code that also includes XPCOM, without trouble. Benoit > > > zw > ______________________________**_________________ > dev-platform mailing list > dev-platform@lists.mozilla.org > https://lists.mozilla.org/**listinfo/dev-platform<https://lists.mozilla.org/listinfo/dev-platform> > _______________________________________________ dev-platform mailing list dev-platform@lists.mozilla.org https://lists.mozilla.org/listinfo/dev-platform
