Re: [xml-dev] XML vs relational database

[Rick Marshall <rjm@zenucom.com>]

I play this game where I turn off the optimiser (I still use C ;) ) and 
profile my code. Then I turn it on and profile the code again.

Over many years I've managed to keep the gap at about 5%.

Why do it? ...

Anyway here's a more interesting take on reflection. It matters to 
Microsoft (and many database engines), but not necessarily to 
Unix/Linux. Why?

Well if you have long lived processes and you can create new modules 
then it's nice to know that the engine/program can somehow work out that 
there might be a better way. In the world of short lived processes 
(typically but not necessarily Unix/Linux) changes/improvements are 
always available.

The paradigm determines the importance of the theory.

Rick

Elliotte Harold wrote:
> Tei wrote:
>
>> 2) As soon runtime optimization make Java faster than C++, some 
>> people will switch. The people that need raw speed.
>
>
> Hindsight's 20/20. That's already happened. :-)
>
> It depends on the problem of course. Some problems are more amenable 
> to compile time optimization and some to runtime optimization. These 
> days sometimes Java is faster and sometimes C++ is faster. For my 
> needs, they;re so close to each other either way, that it hardly matters.
>