[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [MPFRCPP] order of initialization of globals
From: |
Alexey Beshenov |
Subject: |
Re: [MPFRCPP] order of initialization of globals |
Date: |
Sun, 5 Aug 2007 18:32:13 +0400 |
User-agent: |
KMail/1.9.5 |
> You can download Visual C++ 2005 Express Edition for free.
Thank you. I'm working only with GNU/Linux.
>> I've attached a replacement for initialization.cpp.
> It looks like the same sort of change I made, so that should work.
OK. I'll publish new version with corrections.
> One thing about the test applications is that they don't act like unit
> tests -- they act more like compiler tests. In other words, if the
> test application links and runs without crashing, its assumed that the
> code compiled correctly. However, its unclear from looking at the
> output of the test programs if the code is actually working as
> desired.
Yes, that's right. Since MPFRCPP is an interface, there's nothing to test
except possibility of compilation. Numeric part of MPFRCPP depends on GMP /
MPFR. Both GMP and MPFR come with tests. Numerically, there's nothing to test
in MPFRCPP. I think I should add this note in MPFRCPP documentation (I agree
that current 'tests' look confusing).
> For instance, the serialization sample has output that looks confusing
> at first because its printing out the factorial numbers in base 32
> serialization.
It's an example. It uses base 32 since only power of 2 base serialization
works without extra accuracy loss. Yes, it's not human-readable.
> I'm interested in using gmp/mpfr/mpfrcpp for a deep-zooming fractal
> project built using wxWidgets for the UI. I need an arbitrary
> precision math library (preferably with a C++ interface) that can
> compile everyplace that wxWidgets can run, or at least *nix, Windows
> and Mac OS.
MPFRCPP could be compiled with every modern C++ compiler, I think. It is
stable now, but it comes without Automake stuff. I'm very interested in real
projects using MPFRCPP, so, please, contact me with any comments, questions
and suggestions.
--
Alexey Beshenov. http://beshenov.ru/