IUBio GIL .. BIOSCI/Bionet News .. Biosequences .. Software .. FTP

Molbio Software Standard

Reinhard Doelz doelz at comp.bioz.unibas.ch
Fri Apr 2 03:11:20 EST 1993

In article <1993Apr1.215716.5202 at gserv1.dl.ac.uk>, r.nakisa at ic.ac.uk ((Ramin Charles Nakisa) ramin) writes:

|> Here is an outline to discuss:
|> * The source is written in C.  C really is very portable and as an added
|>   bonus it allows us to use Don Gilbert's amazing ureadseq module to
|>   read and write sequences in all the standard formats.  This would also
|>   allow us to build up a standard library of functions, so that we don't
|>   have to re-invent the wheel with every program.

say _ANSI C_ and POSIX system calls. Never use dedicated tricks like 
memory mapping, byte-order dependent routines, and system-specific binary
data setups. 
|> * It should work when compiled for MS-DOS, Windows (quickwin), Macintosh
|>   (simple window format?), UNIX and X.
quite ambitious. You MUST include a text interface, and I would also 
want to have the VMS operating system in there. It would be great if
people would read a style guide to avoid spagetti code. An API below 
the UI needs to be defined and then you can put all you want on top. 

|> * It must be able to read and write all the standard sequence formats.
Why? A reasonabpe input API is sufficient to have all sequences imported
easily. Think about future sequence formats, which would need to be read 

|> * There should be a standard interface.  The most portable interface is
|>   one that consists of only printf and scanf.  A standard graphic
|>   interface would be very difficult to define in a platform-independent
|>   way.  Compiler's can usually produce basic windows for windowing
|>   operating systems eg. the quickwin library on the MS C/C++ 7.0 compiler.
Hoppla. You're putting things upside down if you were talking windows 
up in the definition. A reasonable API with structures defined for 
parameter transmission to the code can well live with complex I/O statements
if it is only modular enough. 

|> * It should offer sensible default parameters when possible, and
|>   use standard command-line switches.
How to use these on a Mac?  :-) 
|> * Graphic output should be in at least two formats: Postscript and HPGL
Again, if a meta language can be agreed on, the output is only a driver
which needs to be plugged in. Make sure that the API on the output side is 
defined and it isn't a problem no longer.

|> * All source should be included and beautified.
This might kill you. Who is really interested in getting source? If it works,
it's fine. No need to fiddle with the compilesr of bersion X.Y.Z, patchlevel
1425 (but make sure that you have the XXX product disbled). Copyright 
procedures become significantly more important, and if there is a reasobable
object library (so that you can link to) I would be satisfied. 

|> If you have ever used particularly good/bad molbio software then here is
|> your chance to have a say before it's written!

You have forgotten the networks. The furure will be much more client/server
computing than you can currently imagine. 

|    Dr. Reinhard Doelz            | RFC     doelz at urz.unibas.ch         |
|      Biocomputing                | DECNET  20579::48130::doelz         |
|Biozentrum der Universitaet       | X25     022846211142036::doelz      |
|   Klingelbergstrasse 70          | FAX     x41 61 261- 6760 or 267- 2078     
|     CH 4056 Basel                | TEL     x41 61 267- 2076 or 2247    |   
+------------- bioftp.unibas.ch is the SWISS EMBnet node ----------------+

More information about the Bio-soft mailing list

Send comments to us at archive@iubioarchive.bio.net