Boost logo

Boost :

From: Daniel Berlin (dan_at_[hidden])
Date: 2000-08-22 12:22:51


On Tue, 22 Aug 2000 mwa_at_[hidden] wrote:

>
> I strongly disagree with this. There are not many 'excellent alternatives'
> if you are building applications for Windows 2000, using COM/ATL/WTL. The
> Intel Compiler is just as broken as the MS one, but in different ways, as
> is the Borland compiler.

Neither of these are true.

Intel's compiler is based on EDG's front end, which is the most standards
compliant around. Borland's compiler is also very compliant, except in a
few library issues.

MS is nowhere near either.
> I've looked at KAI and the last version of
the > MetroWerks toolset, but they are also wanting in platform support.
>
What you mean is that neither particularly care all that much about
windows.
KAI's compiler is available on every single platform i can think of.

> I would prefer to use any number of other platforms for particular jobs -
> but the reality is that I have to support Windows at the expense of
> everything else, and the safest way to do that is to use MS software.

That would be your choice.
>
> I have replaced my standard library with the latest Dinkumware tools, and
> boost offers a valuable set of extensions to that. If I were to lose those,
> I would be sorely disappointed.

Sorely disappointed in the same we we are with MS's non-compliance?
> If they were included in a future version
> of the standard, then there would have to be a MS platform implementation
> for the standard to be anything other than pie-in-the-sky.

An assumption you make here is that MS's platform will stay as valuable as
it is now.
> 'The real world'
> has always been a major consideration for the standards committee (as I
> understand it). Otherwise, we'd have a whole host of changes that would
> have broken existing code-bases. It is sensible and desirable that the
> standard might offer to take a step up from the existing base, but to leave
> it behind entirely could be construed as foolhardy.
>
> Furthermore, MS long-term plans are to achieve full ISO compliance.

Says who?
They seem to care more about adding hteir own extensions than reaching
compliance.
> I
> guess they will reach compliance in about the same time-frame as Borland,
> MetroWerks, GCC et al.

Um, the difference being that the others listed are almost fully
compliant, and are at least usable, whereas MS is nowhere near compliance.
> The specific route they take to get there may be > different, however.
>
> It is also worth bearing in mind that the MS VisualStudio platform has been
> somewhat in flux over the past two years, as they have tried to flesh out
> the strategy which became .NET - tying together solutions to many problems,
> including the biggies:moving to 64-bit and addressing internet
> technologies, without breaking everybody's legacy code.

Which they did anyway.
But that's another story.
And why should boost care about whether MS is having problems coming up
with a coherent strategy?
> I suspect we will
> see more regular releases in the future (like we did in the past), as those
> plans stabilize, and they can assess the take-up on C# and managed C++.
>

Another suspcion of yours.
I used to work for MS. In developer tools. I have no such disillusionment
as to what will happen in the future.


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk