Author Message



We are a heterogeneous group of scientists and students who feel that our  
work is continuously hindered by computer environments dominated by  
incompatible scientific tools and monstrous software packages (too often  
claiming to do everything).  Rather than being able to use different tools  
together in a flexible, interactive system, we find ourselves spending too  
much time converting between different data formats, writing throw-away  
tools and I/O parsers, and worrying about how to get to a particular goal  
rather than what it means to have attained it.

What we need is an object-oriented scientific environment where the tools  
we choose to use are integrated without being parts of a closed system,  
highly interactive, and extendable (both by the addition of our own  
specialized objects and by combining the available tools - graphically).  
The use of such an environment should be a natural extension of our work,  
requiring a very short learning phase and practically no  
user-documentation.  The user should feel encouraged to explore different  
possibilities, testing his/her own scientific ideas without worrying too  
much about whether the system is able to cope (within reason, of course).  
By building an open and object-oriented system, each user should be able  
to draw upon and combine those tools necessary or conducive to a  
particular task: we all need to organize and archive our data, display  
results, and combine information from many different sources.

Currently, the group consists of people from very different fields:  
Molecular Biology, Computer Science, Physics and Astronomy, and Geography.  
While many of our needs for such an environment diverge, the underlying  
motivation is the same: no matter what you want to call it, we need a  
system which helps rather than hinders the organization of our scientific  
data and daily work.


What else is there?  No other system offers the same power, a totally  
object-oriented developers environment, UNIX-compatibility,  
highly-interactive and standardized user-interface.  With the advent of  
NS486 and the expectation that NeXTSTEP will soon appear on a broad range  
of hardware platforms, such a system will soon be nearly universally  
available.  Classical workstation vendors (SUN, IBM, HP, and the rest) now  
have little to offer, no similar tools exist for the standard X-Windows  
GUI, if Apple had a system it would only run on their hardware, and the  
mass-market Windows/DOS world is a developer's nightmare (either despite,  
or more probably because of, OLE).

In order to progress, we need an open discussion of how best to develop  
such an environment.  In NeXTSTEP parlance, do we simply have to put  
together a set of scientific protocols and a common API for our  
"Hyper-knowledge" servers? Should be use distributed objects?  What  
minimum set of tools do we need in order to start working with the  
environment?  What tool-needs do we all have in common and which ones are  
best developed by specialized sub-groups?  How many already existing tools  
can/should be integrated into the system (e.g. Mathematica)?  Do we need  
an "AVS"-like tool for data-flow manipulation?

Mailing list and Anonymous ftp

In order to take part to the project or to push the discussion forewords,  

All (interesting) suggestions, projects and sources will be archive and  
soon available (anonymous ftp).


The European Molecular Biology Laboratory (Heidelberg, Germany) has  
consented to provide the computer resources for a mailing list and  
anonymous ftp services.

Mon, 16 Oct 1995 03:07:35 GMT
 [ 1 post ] 

 Relevant Pages 

Powered by phpBB® Forum Software