[GRASS5] Re: Programming language C vs c++

[Bernhard Reiter]

John,
sorry for not getting back earlier on this,
it was swapped out of focus.

On Tue, Feb 05, 2002 at 10:52:03PM +1100, John Reid wrote:

>  >There are also general doubts regarding
>  >the object-orientated approach in computer science now.
>  >It concerns databases and programming languages.

> Can you please provide further pointers to where I can start to read up
> about this?  

Unfortunatly there are not many papers I can cite
regarding oo and programming languages. The advogato-thread already
mentioned explains the set of feeling that some practicioneers have.
It also make more sense if you know more about the people behind
some of the comments.

Then there was some debate about this on the python newsgroup.
It will not be easy to dig that up. Regularity there are questions
like: How to code design patterns with Pythons. The answers
sometimes lead to debate on beautiful code in general.

It boils down that oo may make program executing harder to follow
and that some languages lend themselfs to huge object-hierachies
which will never be reused often, thus adding unnecessary complexity.

OO still can be good, but it is a way of thinking and only plays out
its advantages when applied to certain problems.

> Most of the concerns that I am aware of in the database world seem to be
> related more to the "mapping" of concepts between the object-oriented
> approach and the object-relational data model.  

I once (end 1999) peeked through a scientific book in the 
library of UWM Milwaukee which pleaded for oo-rdbms and had reasons
with references.

My hypothetis: You have to map all data on a file-system in the end.
Database systems are a lot about optimising data access.
To archive this, you have to use what you know about data.
People might be able to do this better with rdbms.

> Would the concerns you
> refer to be related to what appears to me as a potential problem imposed
> by the necessity to define functions as a member or friend in the class
> definition?  I'm no expert, 

My neither, though I've had some computer-science classes about databases.
Not completley getting where you are aiming at with the question
I tend to answer it negatively.

> It is probably worth keeping in mind for future GRASS development that
> both Geomatics standards I am aware of (ISO and OpenGIS) are based on
> the object-oriented approach.

These standards are more about data structures then about
implementing them. The object-orientated approach is fine there.
As side note: you can nicely develop in an object
orientated way in almost any programming language.

>  >	http://www.advogato.org/article/207.html

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 248 bytes
Desc: not available
Url : http://lists.osgeo.org/pipermail/grass-dev/attachments/20020205/82441a93/attachment.bin