Re: [xsl] Need some Advice

["Wendell Piez wapiez@xxxxxxxxxxxxxxx" <xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx>]

Hi,

To Mike's story, isn't it scary to contemplate how much important code was
migrated in after being written surreptitiously as an effort to solve the
actual problem along with the immediate problem?

Less puzzlingly (maybe), what I am taking away from the conversation is
(again) the importance of unit testing and regression testing. If the OP's
customer had those, they would not have to freak out so hard about
programming style and idiom. At the moment they cannot see the forest for
the Tinytrees.

Cheers, Wendell


On Sat, Feb 5, 2022 at 4:05 AM Michael Kay mike@xxxxxxxxxxxx <
xsl-list-service@xxxxxxxxxxxxxxxxxxxxxx> wrote:

> The resistance is partly political and partly because the
> > code is in production and there are no test cases that anyone can run
> > as regression. It's a million dollar financial services business in
> > the hands of a few developers that have no clue how to code. So there
> > is a fear that something may break and no one knows why!
> >
>
>
> That's a very real problem. I've been in places that didn't dare to change
> the code because it was highly fragile and there weren't enough regression
> tests to be confident that the change wouldn't break things. It's a very
> unpleasant hole to be in, but it's not a problem you can ignore. I remember
> one time where we were handling a badly-designed piece of code with kid
> gloves for about two years before I persuaded the  management that it was
> causing such unreliability that we had to bite the bullet and rewrite it,
> and by that time I had surreptitiously built up a large collection of test
> cases, produced as a side effect of the time spent debugging bug reports
> from the field.
>
> Michael Kay
> Saxonica
> 
>
>

-- 
...Wendell Piez... ...wendell -at- nist -dot- gov...
...wendellpiez.com... ...pellucidliterature.org... ...pausepress.org...
...github.com/wendellpiez... ...gitlab.coko.foundation/wendell...