Fortran - Re: fUnit finally released

Author

Re: fUnit finally released

Arjen Markus

2006-05-22, 8:04 am

For it is worth, I have done a more or less similar job in (mostly)
Fortran -
see http://flibs.sf.net

The really really difficult part is not so much getting supporting
software but
developing the discipline and finding the time and energy to develop
and maintain
the unit tests and anything else ... Oh well, at least one has one less
excuse :)

Regards,

Arjen

Gordon Sande

2006-05-22, 10:02 pm

On 2006-05-22 10:48:31 -0300, "Arjen Markus" <arjen.markus@wldelft.nl> said:

> For it is worth, I have done a more or less similar job in (mostly)
> Fortran -
> see http://flibs.sf.net
>
> The really really difficult part is not so much getting supporting
> software but
> developing the discipline and finding the time and energy to develop
> and maintain
> the unit tests and anything else ... Oh well, at least one has one less
> excuse :)
>
> Regards,
>
> Arjen

I think there is some confusion on what a UNIT is here.

One usage: Unit and integrated testing are part of getting large systems
to work. Here unit means component.

Another usage: Units checking of SI vrs Imperial vrs (bastard* or) engineering
units. Is it 10 kilometers or 6 miles or 5 nautical miles or what to get there?
Really useful unit checking does both the physical units so accelaeration
should be in length divided by time squared as well as the many many variations
on pure numbers. Dollares vrs Pounds vrs Euros in both current and
constant forms.
Staff in FTEs (full time equivalents) and absolute numbers. And on and on...

Certainly one need for unit (component) checking is to get the units (measures)
correct.

The introduction of one of the chapters in Knuth's TAOCP has 30+ variations
on sorting used in one longish sentence.

I thought that fUnit was directed at unit (component) testing.

*They really are called bastard units when they mix various systems. Look
up some measures of insulation effectiveness for strange ones. Joules
(of energy) per square foot (surface area) per inch (of depth) except
it also come in BTUs.

Paul Van Delst

2006-05-22, 10:02 pm

Gordon Sande wrote:
> On 2006-05-22 10:48:31 -0300, "Arjen Markus" <arjen.markus@wldelft.nl>
> said:
>
>
>
> I think there is some confusion on what a UNIT is here.
>
> One usage: Unit and integrated testing are part of getting large systems
> to work. Here unit means component.
>
> Another usage: Units checking of SI vrs Imperial vrs (bastard* or)
> engineering
> units. Is it 10 kilometers or 6 miles or 5 nautical miles or what to get
> there?
> Really useful unit checking does both the physical units so accelaeration
> should be in length divided by time squared as well as the many many
> variations
> on pure numbers. Dollares vrs Pounds vrs Euros in both current and
> constant forms.
> Staff in FTEs (full time equivalents) and absolute numbers. And on and
> on...
>
> Certainly one need for unit (component) checking is to get the units
> (measures)
> correct.
>
> The introduction of one of the chapters in Knuth's TAOCP has 30+ variations
> on sorting used in one longish sentence.
>
> I thought that fUnit was directed at unit (component) testing.

I don't think anyone thinks otherwise (do they?). Where do you see the confusion between
unit (component) and unit (measurement) testing? Was it the use example I quoted regarding
my code that does conversion between different units (measurement)? If so, then my
apologies -- I didn't express myself clearly enough: I meant unit (component) testing of
my unit (measurement) conversion routines.

cheers,

paulv

--
Paul van Delst Ride lots.
CIMSS @ NOAA/NCEP/EMC Eddy Merckx
Ph: (301)763-8000 x7748
Fax:(301)763-8545

Bil Kleb

2006-05-31, 8:10 am

Arjen Markus wrote:
> For it is worth, I have done a more or less similar job in (mostly)
> Fortran - see http://flibs.sf.net
>
> The really really difficult part is not so much getting supporting
> software but developing the discipline and finding the time and
> energy to develop and maintain the unit tests [..]

That's why I try to code test-first whenever possible.
Once having immersed myself in that for about a month, I have found
that I can no longer imagine doing it any other way.

Not having that safety-net of automated tests is downright
scary. Besides creating tight, testable, cohesive, but uncoupled
code, I get other benefits for free like executable documentation/
specification and a custom-tailored debugger. In 6 years of
coding test-first in Ruby, I have never opened its debugger.

As I argue in the /Computational Simulation and the Scientific
Method/ paper[1], I see coding without tests to be a transgression
against the Scientific Method.

Regards,
--
Bil
http://fun3d.larc.nasa.gov

[1] See http://rubyurl.com/qIK

Sponsored Links