[Haskell-cafe] Can Haskell outperform C++?
es at ertes.de
Sat May 19 08:57:38 CEST 2012
wren ng thornton <wren at freegeek.org> wrote:
> However, while the "logical" interpretation of Ertugrul's words may be
> that simple-mindedness implies performance-desire, that interpretation
> is not the only one available to the standard interpretation of his
> words, nor IMO the dominant interpretation. It is equally valid to
> interpret them as saying that the people under discussion are
> simpletons, and that those people desire the best performance
> possible. (I.e., an attributive rather than restrictive reading of the
> adjective.) This latter interpretation is perfectly valid ---as the
> semantics of the utterance---, but is pejorative of the people under
> discussion; and that pejoration is what Ryan was (fairly) calling
> Ertugrul out on.
Well, the meaning was the logical one: Simple-mindedness implies desire
for maximum performance possible. Even that is an overgeneralization.
People can be simple-minded in many ways. I hoped that my point would
Ryan got right that it was indeed an accusation. I did not specify to
what group it was directed, which was probably the reason for the
confusion. The unconditional desire for maximum possible object code
performance is usually very stupid, not to mention impossible to reach
with any high level language and any multi-tasking operating system. To
get the maximum possible performance you must write an ring 0
application in assembly language and boot directly into it.
Haskell delivers reasonable performance for almost all non-embedded
applications, and for the extreme edge cases one can still switch to C
or assembly using the FFI. Haskell's average penalty compared to C is
no reason to write the entire application in C. That was my main point.
nightmare = unsafePerformIO (getWrongWife >>= sex)
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 836 bytes
Desc: not available
More information about the Haskell-Cafe