Re: [boost] GCC 2.95.2 and <limits>

From: wb_at_[hidden]
Date: 2000-10-15 16:57:06

I have a <limits> replacement that we have been (lightly) using with
gcc 2.95.2, and have just received permission to make it available to
boost.org.

Before coding this version, we took a look at the (then-beta) STLport
4.0 version, and noted that it was not strictly a header:  there was a
library component that seemed intended for separate compilation.  Our
preference was for a header-only version, which I then created.

The code (of course) comes with no guarantees, especially since
<limits> is, by its nature, inherently platform-dependent.  (The
STLport version we looked at contained a comparable caveat).  Although
we did no ultra-heavy stress-testing on it, our header has worked for
us quite adequately on several platforms.  There are a few embedded
macro usages which will need definitions, but they ought be fairly
straightforward.  These are things like, e.g.,

    #define  NOTHROW  throw()
    #define  BEGIN_NAMESPACE(ns)  namespace ns  {
    #define  BEGIN_UNNAMED_NAMESPACE  namespace  {

I will be pleased to correspond privately regarding this header, and
(assuming there is interest) would like to know where to send it (~600+
lines) so as to avoid polluting the boost mailing list.

Best,

        - WEB

David Abrahams <abrahams_at_[hidden]> wrote on Sun Oct 15 15:32:12 2000:

| Well, hijacking <limits> from STLport in the meantime should be easy enough,
| should someone desire to do it ;-)
|
| ----- Original Message -----
| From: "Beman Dawes" <beman_at_[hidden]>
| To: "Boost.org mailing list" <boost_at_[hidden]>
| Sent: Sunday, October 15, 2000 12:50 PM
| Subject: [boost] GCC 2.95.2 and <limits>
|
| > There has been some discussion of supplying a boost <limits> headers so
| GCC
| > 2.95.2 will pass more of the boost regression tests.
| >
| > The folks at http://gcc.gnu.org/ are now saying they hope for a release of
| > GCC 3.0 "by the end of the year".
| >
| > So rather than get on the <limits> slippery slope, let's just wait for GCC
| > 3.0.