Re: [MLUG] coding languages used in genetics

[Jonathan King <EMAIL:PROTECTED>]

On Thu, 1 Apr 2004, Mike Miller wrote:

> Might as well post this.  I just did a search of the long list of programs
> given here:  http://linkage.rockefeller.edu/soft/
> Most programs should be listed there.  They have a line for "source code
> language" so I grepped that line to get a list of languages used in coding
> these programs.  Then I grepped that list and came up with these data:

Interesting.  I think it's worth noting, though, that you would have 
gotten very different numbers if you'd looked into code that wasn't 
doing linkage analysis.  In the sequence world, stuff lke EMBOSS and 
the bioperl project are huge, and there ain't no FORTRAN or Pascal 
there. :-)
 
In any case, I think it's worth combining C and C++, Pascal and 
Delphi, S and R and re-tabulating:

231 total
112 C/C++
 49 None Listed
 32 FORTRAN
 30 Pascal/Delphi
  9 S/S-plus/R  
  7 Java
  7 Perl
  4 Basic
  2 SAS
  1 Python

So the languages on this list that most MLUGers are probably less
familiar with are S-plus/R and SAS.  SAS is the language of big-time
statistical data analysis.  I think most of you would think SAS
was...pretty awful.  R is an open-source implementation of S-plus,
and is actually quite an interesting system. S(-plus) was written
for stats geeks by computer geeks at Bell Labs, so it's actually a 
decently designed language, and then the open-source R people have 
been improving it even more.

> A few years ago, someone took one of the field's most widely used
> Pascal programs, ran it through p2c, compiled with gcc and managed
> to get it to run.  It was "six to 30 times faster" than when
> compiled with Pascal.  I think this means that the Pascal
> compilers suck compared to gcc.  They had made no changes to the
> algorithms in the code.

Without even seeing the code, I'm willing to bet this happened
because of Pascal's boundary checks on arrays.  Once you start
nesting loops and cruising over multi-dim. arrays, that penalty
probably really adds up.  Back in the day, this was one big
advantage of C that people used to club the Pascal community to
death.  The other was strings.

That said, the next time somebody ruins your day by exploiting a 
buffer-overflow bug, you can make it even worse by imagining some 
Pascal weenie smirking at you saying "See, we told you so".

jking


_______________________________________________
members mailing list
EMAIL:PROTECTED
http://mlug.missouri.edu/mailman/listinfo/members