From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bob McWhirter To: egcs@egcs.cygnus.com Subject: Re: Bug in libm or libstdc++. Date: Wed, 24 Feb 1999 15:42:00 -0000 Message-id: In-reply-to: < 3.0.6.32.19990222090333.00847df0@pop.globalserve.net > References: <3.0.6.32.19990222090333.00847df0@pop.globalserve.net> X-SW-Source: 1999-02/msg01221.html On Mon, 22 Feb 1999, Paul Derbyshire wrote: > C:\PGD\C++\omega\ofc>gcc ntst.cc -o ntst.exe -Wall -lm -lstdcxx > c:/djgpp/tmp/cc5PNfut.o(.text+0x51):ntst.cc: undefined reference to > `atan(long double)' [ snip ] > What the hell? Is that -really- necessary? Pipe down, already. > Stroustrup 3rd Ed definitely informs me that long double overloads for > these and other functions are supposed to be in either libm or libstdc++. Uhhh... if you're referring to page 660 (section 22.3) it most certainly does not make any such guarantee with regards to -long double- datatypes. > Using egcs 1.1.1 and the libm and libstdc++ that came with it. at least on my box (sparc solaris), libm is vendor provided. gnu/include/g++/cmath -does- make reference to the math functions that use long doubles for arguments, but I don't think any library on my box actually supports them. Possibly there for platforms that do? There for platforms that typedef a long double to be the same as a double? (nm'ing vendor's libm and stdc++ does not reveal math functions taking long doubles, but possibly some other lib I'm ignoring? I did find math functions take complex objects as arguments though.) -Bob From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bob McWhirter To: egcs@egcs.cygnus.com Subject: Re: Bug in libm or libstdc++. Date: Sun, 28 Feb 1999 22:53:00 -0000 Message-ID: References: <3.0.6.32.19990222090333.00847df0@pop.globalserve.net> X-SW-Source: 1999-02n/msg01230.html Message-ID: <19990228225300.K2g8Nrffl-JUpEZ-NzgXpBUe_LhR5wmMmQfKVaslOFQ@z> On Mon, 22 Feb 1999, Paul Derbyshire wrote: > C:\PGD\C++\omega\ofc>gcc ntst.cc -o ntst.exe -Wall -lm -lstdcxx > c:/djgpp/tmp/cc5PNfut.o(.text+0x51):ntst.cc: undefined reference to > `atan(long double)' [ snip ] > What the hell? Is that -really- necessary? Pipe down, already. > Stroustrup 3rd Ed definitely informs me that long double overloads for > these and other functions are supposed to be in either libm or libstdc++. Uhhh... if you're referring to page 660 (section 22.3) it most certainly does not make any such guarantee with regards to -long double- datatypes. > Using egcs 1.1.1 and the libm and libstdc++ that came with it. at least on my box (sparc solaris), libm is vendor provided. gnu/include/g++/cmath -does- make reference to the math functions that use long doubles for arguments, but I don't think any library on my box actually supports them. Possibly there for platforms that do? There for platforms that typedef a long double to be the same as a double? (nm'ing vendor's libm and stdc++ does not reveal math functions taking long doubles, but possibly some other lib I'm ignoring? I did find math functions take complex objects as arguments though.) -Bob