On Tue, 2006-12-12 at 04:33 -0800, llothar wrote:
> Hey there we
> still have a discussion about "semantics" of the threading
> implementation for eiffel. Hello this is >15 years under discussion
> now. No nobody seems to realize that this is a simple proof that they
> failed. Point.
So far, yes. Does this mean that the Liskov/Wing programming
model is flawed, inconsistent, or not working? No. Could you point
to what exactly has/have been the point(s) of failure of Eiffel
concurrency? That is, why do you think it cannot work, not why
"separate" isn't currently working, or not working portably, or
not on all platforms, or not with your preferred compiler?
Rather, can you show us a conceptual mistake?
(I am prepared for a comment like "well, if the fact that it
doesn't work doesn't convince you...", but these comments have
been made when run time dispatching was considered and was=20
dismissed, so that there was no inheritance based OO in Ada 79/83.
It was later added when people had learned you can do this
efficiently, even in real time systems. (About a dozen years
later, inheritance based OO was added to the language standard.))
> Since the release of ECMA Eiffel everything is getting worse. No
> Halstenbach, No Visual Eiffel, No SmartEiffel anymore (they are now
> more and more moving to there own Loria-Eiffel).
If these business have failed this fact can be measured against the
competition (among other things): There is now hardly any
competition WRT tools for PC desktop apps. This market is
"cleverly" controlled and watched by one company, covering
more than 90 % of everything PC. No surprise.
In the PC market, any PC language + libraries that isn't marketed
by Microsoft is best compared to another language not marketed by
Microsoft. Delphi? Intel C++? Cobol? F95? GCC? What is their share of
the PC tools market? Why did Borland Pascal "die"?
In the embedded Market, what is the share of C# applications?
VB in car electronics? OTOH, verification techniques using
assertions seem to be on the rise in some embedded systems
markets. There is real money, if not yours or mine.
Is C# strong in banking? Do they use Exchange servers and
VB in order to process transactions through their back end
machinery? Have insurance companies abandoned array languages
in favor of those other immensely successful
alive-for-as-long-as-MS-can-sell-them-profitably languages
(MS Fortran died when...)?
The Liskov/Wing subtyping that could be made fully present in
Eiffel is a dozen years old, yes, and it is not yet implemented
either. Yes.
The only difficulty I see here is that Eiffel does try to provide=20
Liskov/Wing subtypes while the marketing of other tool suppliers
just talks around the issue. They give you patchwork software and
patchwork languages that are bound to change soon so that
you have an opportunity to buy another license from them because
of language changes, whether you need language changes or not.
This has another advantage: If their compilers don't compile
the envisioned language, they can withhold language features,
and then the compiler will compile the currently envisioned
language.
The fact that Loria does what a research institute is entitled
to do is probably missing some of the Eiffel engineering incentives.
But that is what research can do if there is still funding even=20
though they concentrate on only a selection of incentives.
As long as their work is not guided =C2=AB par l'aide =C3=A0 la
cr=C3=A9ati=
on
d'entreprises =C2=BB, and as long as Eiffel enterprises see little
advantage in concurrent execution, why should they act differently?
Though seen from the quote above, maybe you know someone in France
who can address your comments to the ministry of education in Paris?
> I don't believe
> the "open source the code and you will find people helping you" mantra.
And I don't believe people whining about failed businesses
when they want to explain to me how the world works in all
cases. Remember that every little language competes with C#, VB,
Cobol, Java, and their respective monetary and social funding.
In niches. The Eiffel market is small. Eiffel syntax is boring.
So there.
|
