[lsb-discuss] LSB 5.0 Beta 1

[Dallman, John]

R P Herrold [mailto:herrold at owlriver.com] wrote:

> Tactically we will probably end up fixing as breakage is noted,
> but emitting headers out of the database has proven to be
> challenging, and so we will not get to a final solution in 5.0

Oh, well.

> Obviously GCC is not the be all and end of all of C++ compilers.
> The LLVM / Clang suite is quite credible and useful in local testing
> in ferretting out assumptions in code not supported in the C++
> standards

Indeed. But among the uses my employer makes of C++ is shipping closed-
source libraries to ISVs who need to integrate them into their
applications. Using Clang for that on Linux would add considerable fear,
uncertainty and doubt to the customers' integration task. Yes, it should
all work fine, but proving that to all the customers, via support people
who think in Windows, would be hard work. We use Clang on OS X and we're
happy with it there, but OS X is being shifted to Clang by Apple, without
much option for ISVs.

The reason we use LSB is that it means those libraries can currently run
on more or less any Linux. Obviously code built to an LSB standard with
support for C++11 would need a C++11 capable system to run on, but those
should become fairly standard in a couple of years' time, now that RHEL7
is in beta. So we're going to have to wait for a later LSB.

> What features are in your Requirements that you hope to see
> expressed which are new in C++0X^h^h11 ?

Sadly, different product teams have different C++11 features that they're
interested in, ranging from "nothing" to "almost everything". The ones that
are most interested in C++11 don't currently produce LSB-compliant code
anyway, and the lack of support for C++11 just means they become less
interested in the idea.

thanks,

--
John Dallman

-----Original Message-----
From: R P Herrold [mailto:herrold at owlriver.com]
Sent: Wednesday, December 11, 2013 8:04 PM
To: Dallman, John
Cc: lsb-discuss
Subject: LSB 5.0 Beta 1

On Wed, 11 Dec 2013, Dallman, John wrote:

> > The LSB workgroup is pleased to announce the availability of LSB 5.0
> > Beta 1.
>
> How much support for C++11 does this have? Sp4cifically, GCC 4.7.2 and
> later?

We discussed this in the call today, as well in prior calls.

The relevant parts of the minutes today are:

C++11 discussion -- again, the upstream standard is itself in
motion, and might fairly be characterized as shipped partially baked, such that GCC and others are chasing a moving target.
Tactically we will probably end up fixing as breakage is noted, but emitting headers out of the database has proven to be challenging, and so we will not get to a final solution in
5.0

===============

Obviously GCC is not the be all and end of all of C++ compilers.  The LLVM / Clang suite is quite credible and useful in local testing in ferretting out assumptions in code not supported in the C++ standards

===============

What features are in your Requirements that you hope to see expressed which are new in C++0X^h^h11 ?

( Stroustrup discusses the partial completeness yet ratification in his typical low-keyed manner in:
        http://www.stroustrup.com/C++11FAQ.html#1x

The next two sections of that webpage merit contemplation in this context -- heck, the whole page is worth reading and thinking about carefully )

-- Russ herrold
-----------------
Siemens Industry Software Limited is a limited company registered in England and Wales.
Registered number: 3476850.
Registered office: Faraday House, Sir William Siemens Square, Frimley, Surrey, GU16 8QD.