Re: Cross-Platform Windows plus Mac

In article <1194542237.815877@[EMAIL PROTECTED]
>,
 Michael Ash <mike@[EMAIL PROTECTED]
> wrote:

> In comp.sys.mac.programmer.help Ben Artin <macdev@[EMAIL PROTECTED]
> wrote:
> > In article <1194537879.688818@[EMAIL PROTECTED]
>, Michael Ash 
> > <mike@[EMAIL PROTECTED]
> wrote:
> > 
> >> In comp.sys.mac.programmer.help Ben Artin <macdev@[EMAIL PROTECTED]
> wrote:
> >> > Given a language feature that is semantically equivalent in two 
> >> > languages (such as exceptions), not being able to use that feature 
> >> > across the two languages has nothing to do with proper separation
and 
> >> > everything to do with running into a limitation of the language
bridge 
> >> > and sucking it up.
> >> 
> >> You could replace "exceptions" with "objects" in the above and still
have 
> >> an accurate statement.
> >> 
> >> Of course this isn't really a defense of ObjC++, more like a proof
that it 
> >> *must* suck, no matter how it's done.
> > 
> > Well, yes and no. 
> 
> You say this, but then you go on to say what appears to be "yes and
yes". 
> :)

I guess you can look at it that way :-)

> > This line of reasoning ultimately leads you to realize that what you
really 
> > need is a common language runtime with well-defined semantics, and
then you 
> > need all the languages that you care about to work on top of that
runtime. 
> > This is what Microsoft did with CLR, and this is increasingly the
direction 
> > that Apple is going in by bridging the Obj-C runtime to Python, Perl,
Ruby, 
> > etc.
> > 
> > Then you realize that C++ can't work on top of any sane runtime, at
which 
> > point, if you are Microsoft, you invent a modified version of C++ that
can 
> > or, if you are me, you realize that you are better off using a
language 
> > (such as Python) that works on top of a sane runtime.
> 
> In other words, the only way for ObjC++ not to suck would be to do
things 
> that result in no longer having C++ anymore. Ergo, ObjC++ must suck.

What I am trying to say is that by its nature Obj-C++ must suck as much as
C++ 
does, but it doesn't have to add suck of its own. Effectively prohibiting
use of 
exceptions in C++ APIs called by Obj-C is one case in which I think
Obj-C++ 
added suck on top of the suck of C++ (and I was glad to see that's fixed
in the 
64-bit runtime).

Programming in C++ is already like being in a straightjacket. In some
cases, 
Obj-C++ also ties your shoelaces together.

[snip]

> Of course a major consequence of this would be to suddenly make C++
*slower* 
> than the others, due to all the extra indirection. That might be enough
to 
> qualify as "suck", but I'm not sure.

Well, it would to C++ weenies. It cracks me up every time I see a
discussion of 
C++ libraries that goes along the lines of

"We should have a C++ Unicode string class"
"Yeah, but what encoding would it use?"
"Store the encoding in a private member, and convert to other encodings if

needed"
"That would be too slow"
"Never mind then."

Ben

-- 
If this message helped you, consider buying an item
from my wish list: <http://artins.org/ben/wishlist>

I changed my name: <http://periodic-kingdom.org/People/NameChange.php>