LDBL_MIN et al need a test

• Key: STDCXX-10
• Type: Sub-task
• Status: Open
• Priority: Minor
• Assignee: Unassigned
• Reporter: Martin Sebor
• Component/s: Tests
• Affects Version/s: 4.2.0, 4.2.1
• Fix Version/s: 4.2.2

Description

A test needs to be written for STDCXX-8 (and more generally, for the rest of <float.h> and <cfloat> macros). Here's what it might look like:
http://mail-archives.apache.org/mod_mbox/incubator-stdcxx-dev/200508.mbox/[e-mail]%3e

Comments

[Martin Sebor]

Lowered Priority and rescheduled for 4.2.1.

[Farid Zaripov]

I'm not sure that I understand the problem in STDCXX-8.
The FLOAT.cpp config test print's LDBL_MIN macro value as #define _RWSTD_LDBL_MIN %LDBL_MIN% (%LDBL_MIN% is the numerical value of the corresponding macro). Then in include/ansi/float.h we #undefine LDBL_MIN and #define LDBL_MIN as _RWSTD_LDBL_MIN for EDG compiler or not redefine LDBL_MIN for other compilers. So what the difference between LDBL_MIN and _RWSTD_LDBL_MIN? And how _RWSTD_LDBL_MIN macro (which is copy of LDBL_MIN) caused floating point underflow on icc 8.1?

[Martin Sebor]

The block in float.h is only for EDG eccp on Linux (glibc), not for Intel C++.

If I had to guess, the warning pointed out in STDCXX-10 (and similarly in STDCXX-565) is because of some small difference between the value of _RWSTD_LDBL_MIN printed out by FLOAT.h and what the compiler considers to be LDBL_MIN.

[Farid Zaripov]

Defer the issue to the next release.