Software Discipline in Molecular Biology

doelz at urz.unibas.ch doelz at urz.unibas.ch
Sat Jan 18 14:01:26 EST 1992


( Citation from article <1992Jan17.182355.21103 at cica.indiana.edu>, 
gilbertd at ogre.ogre.cica.indiana.edu (Don Gilbert) )
 

Recently, Don Gilbert wrote a very objective, comprehensive article on the 
usage of Macintoshes in the Unix world , and I agree with nearly everything. 
There's just a small BUT there: 

> The drawbacks to relying on A/UX as a Unix platform seem to be:
> 
> (a) It isn't a Sun Sparcstation (this is a drawback for many unix 
>     platforms :).

...

> to 50 MIPS.  These include computers like Sun Sparcstation IPX,
> Silicon Graphics Iris Indigo, and workstations from DEC, HP and
> others.

Without disregarding the success/failure of particular vendors to 
contribute enough 'standard' 'posix' or whatever unix, I think that 
it is well possible to write software in portable fashion. Just look 
at GCG's package which is now available on more or less the same 
source code servicing VAX/VMS, DEC/ULTRIX, SUN/SunOS, and SGI/IRIX. 
The only problem is usually that the developers of molbio software rarely 
have the resources to develop portable code because of 

*) lacking hardware (which lab is daring enough to get everything available
       on the market?) 
*) lacking need (why write a package for a platform which isn't standard
       in the particular lab?) 
*) lacking experience (which grad student can afford to learn all 
       peculiarities of a particular window manager or stdio library ? )

Now as the experience shows many don't like to stay with the commercial
packages only and fiddle around with own hacks. What a mess! Besides 
the horrible effect that once the student got its PHD and leaves the 
lab he/she usually leaves the maintenance of it behind as well (causing
hedaches and sleepless nights for all those who want to do something 
with it afterwards), the style of this software is usually very 
'spontaneous' (spagetti code, and no conformance to interface design rules, 
to name a few).

Therefore I strongly suggest to realize the following: 

1) We need an education pathway which gets us a degree in 'biocomputing' so 
   	that the poor chaps doing software development in biology can cite 
	this as professional experience afterwards. 
2) We should use MOTIF as the evolving standard and also less complicated 
	code for stdio. If you can't make software written for a particular 
	platform run on others, windowing and io are usually the problem. 
3) We should encourage young students to write DOCUMENTATION to their code. 
	Once you ftp something you usually get what you deserve and this is 
	not much. Lacking documentation is the reason that ports are 
	not easy. 
4) If you do a port you should be willing to collaborate with the original 
	author in order to work on the code. The next release whould not 
	be as specific as the earlier one, and others should not need to 
	reinvent the wheel. 


Regards 
Reinhard    


************************************************************************
    Dr. Reinhard Doelz            *     EAN     doelz at urz.unibas.ch
      Biocomputing                *     DECNET  20579::48130::doelz
Biozentrum der Universitaet       *     X25     psi%022846211142::embnet
   Klingelbergstrasse 70          *     FAX     x41 61 261- 6760 
     CH 4056 Basel                *     TEL     x41 61 267- 2076 or 2247   
************************************************************************


(I have a variety of UNIX platforms available, and investigated whether 
to upgrade my mac or get a workstation. Finally, I got an Indigo SGI for 
various reasons and I'm very impressed.)




More information about the Bio-soft mailing list