Re: GNAT 3.11 worth installing

[dewar@gnat.com]

In article <36DE4874.9FA45B71@frqnet.de>,
  Andreas Winckler <andreas.winckler@frqnet.de> wrote:

> With GNAT 3.11 the software crashes uncontrolled with a
> segmentation fault (core dump).

Whether this means that GNAT 3.11 has a problem or not is
of course undetermined.

We sometimes experience situations in which a piece of
faulty code happened to work before, and does not work with
a new version, and of course the assumption is that
something is wrong with the new version of the compiler.

But this is not always the case. FOr example, in one
situation, the code contained a hard coded constant for a
delay statement, which had been tuned for the particular
behavior of a particular version of the compiler. With the
new version of the compiler implementing zero cost
exceptions, this constant was quite wrong, and the result
was that one task kept issuing the delay over and over
again, hogging the CPU!

Yet another example of the mantra "Just because it works,
does not mean it is correct!"

(and the converse of course in this case is "just because
it does not work does not mean there is a compiler bug" :-)

Of course this certainly *could be* a problem in GNAT, but
without more information it is impossible to say.

Speaking for our customers, they have found that 3.11 is
in general far more stable than 3.10, and it contains many
new useful capabilities (see features file in release).

Robert Dewar
Ada Core Technologies

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own