Re: Is that a joke ?

<jabailo@xxxxxxxxxx> wrote in message
news:fOGdnaypLLnQZSPfRVn-2w@xxxxxxxxxxxxxxxx
> 'Slow' is also a word that you should never use if you want any respect
> from
> engineers. Can you quantify what 'slow' means for you? If anything,
> some of my running battles with certain c++ programmers show that it's
> just-as-fast to nearly-as-fast for many key processes such as word search
> ( using Regex ) and highly threaded applications.

I was talking about applications like cad, cam, engineering stuf, math,
scientific applications, data aquisition etc. In that case I hope you
understand that .NET regex performance doesn't help a lot.
Last two years, I've ported a lot of code from C++ to C# and managed C++.
I've seen up to 4x performance decrease and more than 7x resources usage.
Please don't ask me what was the comparison results when tested data
aquisition & analysis, testing just peaked to 100% of CPU usage and the
application was not able to poll data.
Not to mention user interface slowness, probably that will be changed in the
future.

> Performance always matters. Somehow, you have this outdated notion of
> what
> performance in business or Enterprise means. In my business we apply
> strict speed and performance metrics to all our Operations from loading a
> file to updating a database. My motto is "under 1 ms' -- the goal of
> atomic operations should always be less than a millesecond.

Do you really think that this kind of performance is all we need today ?
I suppose that in your business, you still use programs like imaging, math
or statistics, archivers, spread***, utilities etc.
Can you tell me how database performance can help Photoshop, for example,
rendering an image or an archiver to compress files or even the OS itself to
do other usefull tasks ?

>
> What are your applications?
>

I'm working on engeeneering, industrial automation and CAD projects.


> Safe from self-created memory leaks. But there is always the unsafe
> approach if you have the skills.

AFAIK there is no "self-created" leaks.
If a developer use to allocate memory and forgets to free it then don't
blame Win API for that.
What do you think this kind of developers will do in .NET? Maybe GC will
protect framework from their faults but I cannot see how the system will be
protected from their poor coding.


> If you read the papers, the idea of making .NET as the only interface to
> the
> new Longhorn API has been dismissed.

I've read the papers and the only thing I understood is that WinFX will be
the primary programming API. And as you know WinFX is managed code. Of
course Win32/64 will be there too, although it will be called "legacy".

>> 8. Longhorn drivers, kernel and anything "low-level" will be written in
>> native code. LH must run in 64bit processors too, so native code must be
>> natively compiled in 64bit API calls. That means Win32 will become Win64.
>> Why MS don't simply implement WinFX as an OO frontend to Win32/64 API,
>> leaving out the VM?
>
> That's the first brilliant thing you've said.

I'm glad you agreed with that.


>
> Okay, then you must have higher level contacts than this newsgroup...or
> don't you?
>

What you mean by that ? If you asked if I know MS people then the answer is
"NO" with the exception of a few of them who were in the same univ with me
years ago.
Even if I knew Bill Gates, do you think that I could make him change
Microsoft's roadmap ?
One think I've learned all these years, is that nobody, including mr Gates,
makes a decission based on his own will. I'm sure that MS marketing people
either have plans for web service market or believe that personal computer
market is dead.


I wonder what will happen when MS try to rewrite Excel or Word in managed
code (!) Unless they have plans to make Office available as a web service on
subscription basis.

What a future for the "computer science" world !

Regards

Ted Nicols.


.