On 2016-10-15 20:59, Roland Hughes wrote:
> On 09/28/2016 12:25 PM, interest-requ...@qt-project.org wrote:
>
> Don't you have unit tests?
>
> Yes. But which is better, to be forced to use an inherently error-prone
> language (JavaScript) and rely on unit tests to clean up the mess, or to
> use a robust modern language (C++) and have less bugs to find and fix in
> the first place?
When you work off nothing but stories you are hacking on the fly and if
your coders
aren't system architect level people, not just one, but each and every
one of them,
you end in failure. You end up with a developer choosing to store data
in JSON files
for a device taking dozens, some times hundreds of readings per second,
appending
the new reading to the end of a JSON array and writing the entire file
back to an
SD card. Without the event->message->device response->message->event
life cycle
completely mapped out in a solid document you end up with a dozen
programmers
working from a dozen different stories doing it a dozen different ways
so eventually
you end up with an embedded system slamming hundreds of requests onto a
message bus
for data it only needed to get once at startup.
Documents won't avoid that... at my last job we did 'v-model
development' and created hundreds of pages of documentation and then
went off and coded for months ignoring the documentation because it was
bad and never kept up to date. It had quadruplicated definitions of
features that all disagreed. Copied/pasted/modified definitions done
pages at a time. One feature definition stated, if you bothered to
actually read it, that the entire system must not move (hint: its
purpose was to move...). That was one of the multiply copied items. :]
Anyways, any development methodology will fail if implemented poorly.
If the buzz-word is agile, it will likely fail because management won't
be willing to handle it appropriately. Same goes for v-model.
Cheers,
-kyle
_______________________________________________
Interest mailing list
Interest@qt-project.org
http://lists.qt-project.org/mailman/listinfo/interest
