I have solved many problems where the resulting code has been used
for years - without anything like the "rigorous testing" that "should
have been done"...
Many success stories about J/APL come from people solving real
problems and the solution running for years while the "official
approach" was being developed (and tested). I have seen cases where
the official, tested system fails to work as well as the shot from
the hip J/APL solution. I think many people in these forums have had
similar experiences.
IMHO, when a J/APL solution solves a problem, chances of pathological
cases causing bad things to happen are remarkably less than when
using "conventional" languages. This is contributed to by the
conciseness and rigor of "Iverson thinking".
Let me add that I do very much admire the approach to testing that
you have developed for j releases. The paucity of bugs indicates the
high quality of your work.
- joey
At 10:10 -0700 2009/09/26, Roger Hui wrote:
>I think people have used that kind of argument
>against the use of J/APL. ("You have to
>be really smart to use J/APL ...")
>
>Also, I am not smart enough to write shippable
>code that I haven't tested.
>
>
>
>----- Original Message -----
>From: Joey K Tuttle <[email protected]>
>Date: Saturday, September 26, 2009 10:00
>Subject: Re: [Jchat] Is J Cheating
>To: Chat forum <[email protected]>
>
>> A practical point of view -
>>
>> http://www.joelonsoftware.com/items/2009/09/23.html
>>
>> I would say that J offers really good tools (Duct tape and WD40
> > if you will...)
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
