Subject Re: [IBO] ib_query ignores my data!
Author staff@belding
Hello Geoff & others

----- Original Message -----
From: Geoff Worboys <geoff@...>
To: <>
Sent: Friday, March 23, 2001 12:09 PM
Subject: Re: [IBO] ib_query ignores my data!
> I also suggest that this NOT be changed in IBO v3.x. The potential
> for breaking existing applications is too great. Lets leave whatever
> change is decided for IBOv4.

Agree Geoff, changes which require rework should be postponed till a major
revision. However, a change to a default property value will not affect your
code will it? Won't your code continue with the old default value just as if
you selected it the last time you changed it?

> > Because of this ease, I would like to move more towards
> > mathematical correctness, written differently "don't touch
> > my data!".
> I would only ask that you remember back to when right trimming was
> enabled. It seems to me that there were many more questions and
> problems when this was not being performed. I am afraid that the
> arguments for not trimming have so far only convinced me that the
> number of instances where trailing whitespace is significant is very
> small.

I guess you are correct about the frequency of critical whitespace

My software experience is with RT signal processing and control in fighter
radar sets, diagnostic software, maintenance software and document
production. In no case would I want to have component defaults which removed
data, whitespace included.

If we take the conversation away from whitespace for a momement and look at
the general principles, for convenience working with a framework of Data -
User - Developer - Component Developer.
1. Users insert and own the data
2. Developers may or may not know apriori data format
3. Component Developers only rarely will know data formats in specialised
cases and IBO is not for specialists but for IB generalists.

My design thinking is that default behaviour at the component-developer
level which removes data is wrong. In many cases it will be benign, but in
the case where a developer does not realise what the default behaviour does
and gets caught it could give IBO a "quirky" reputation. Even a majority of
benign examples does not overcome the basic flaw in this "being helpful but
possible breaking software" thinking.

If one favors the position to have default trimming and developers can
change the default trimming behaviour then the IBO documentation should go
to extra efforts to include warnings.

I can live with whichever option Jason selects, and have tried to counter
the trimming option for a basic principle.

Best wishes