Re: Ada Shootout program for K-Nucleotide (patches)

[Isaac Gouy <igouy2@yahoo.com>]

On Aug 6, 3:49 am, "Dmitry A. Kazakov" <mail...@dmitry-kazakov.de>
wrote:
> On Thu, 06 Aug 2009 12:34:14 +0200, Georg Bauhaus wrote:
> > Dmitry A. Kazakov schrieb:
> >> On Thu, 06 Aug 2009 10:36:38 +0200, Georg Bauhaus wrote:
>
> >>> Does it matter?  Gives C programmers the illusion that
> >>> it is their language that makes I/O fast when it is
> >>> Unix I/O and the wrappers that have naturally made it
> >>> into its handy byproduct C.
>
> >> Huh, Unix I/O is fast?
>
> > OK, I meant that most programs in the game use simple
> > I/O functions for line output, functions in particular that
> > were developed with Unix. (Calling fputs, or even routines
> > that have read(2) under the hood, I imagine.)
>
> > No one uses fprintf for line output; we have to, if
> > we stick to Text_IO.   If all goes well, Streams.Write
> > will do something close to what everyone else does, and
> > in plain Ada.
>
> These shoot-outs are usually poorly thought and stated. Clearly I/O has
> nothing to do with the language performance when the problem is not
> directly related to communication with the file system. Otherwise, it is
> usually the file system which is a problem and not the language... (:-))
>
> OK, this is not a joke. In fact, Ada.Text_IO is fast enough for its
> purpose, i.e. dealing with *text* files. If we have MBytes of *text*, there
> must be something wrong with us, if we are using crippled OSes, which do
> not distinguish texts and data. Ada was designed in the times when crippled
> OSes didn't dominate the market...
>
> --
> Regards,
> Dmitry A. Kazakovhttp://www.dmitry-kazakov.de


Criticisms of these shoot-outs are usually poorly thought and stated.

If we have MBytes of *text*, we must be doing bioinformatics, if we
are using crippled OSes, which do not distinguish texts and data, we
must be concerned with the present rather than some bygone age.