Template Metaprogramming Made Easy (Huh?)

Nick Sabalausky a at a.a
Thu Sep 10 18:15:25 PDT 2009


"dsimcha" <dsimcha at yahoo.com> wrote in message 
news:h8c52o$2a8e$1 at digitalmars.com...
> == Quote from Nick Sabalausky (a at a.a)'s article
>> And before I get the inevitable "D00d thats soo old U shud by a new 1!",
>> yes, I *could* go buy a new system. But why should I? I don't do a single
>> thing that can't be done just fine on my single-cores. And the only 
>> things
>> that run poorly are the things are written by teenage lazy hack "I don't
>> care about intelligent coding, because everyone should be just like me 
>> and
>> want to sink all their money into new hardware just because they can!"
>
> Not sure I buy this.  Let's analyze it in simple microeconomics.  Both 
> programmer
> time and computer hardware are scarce, expensive commodities.  To some 
> extent, one
> can be substituted for the other.  (A programmer can either spend less 
> time
> writing crappier code that needs more hardware or vice-versa.)  All else 
> being
> equal, you want the cheapest software you can get.
>
> For the sake of this argument, I'm going to assume that the software is 
> paid for
> directly by the consumer, though the argument could be extended to cases 
> where it
> is paid for indirectly (business websites, etc.) and free software.  A 
> company can
> either deliver really unoptimized software for little programmer time, and 
> thus
> cheaply, or really fast software expensively.  As a consumer, you only 
> care about
> *total* cost.  Therefore, as the cost of better hardware goes down, the 
> only
> rational thing to do is spend less time optimizing software.
>
> Of course, this doesn't work for special purpose computers that only run 
> one piece
> of software, but let's say the average computer user runs ~20 pieces of 
> software
> regularly.  If a new computer costs $400, and each piece of software can 
> be made
> on average $20 cheaper by not optimizing it, then you break even.

There are a *lot* of 'if's and assumptions in that analysis.

In general though, I find the "programmer time is more expensive than 
hardware" line to largely be a cop-out.





More information about the Digitalmars-d mailing list