"[H]omer" <spam@xxxxxxx> writes:
> Roy Schestowitz wrote:
>
>> I once knew a guy who never used/s Windows. He was of a foreign
>> origin and he went to university with me. At the time, his computer
>> knowledge seemed totally outlandish. That's how you come to judge a
>> 18-year-old who, at the year of Y2K bugs, does not even use
>> Windows. Click to focus was very unnatural to him, for instance. But
>> guess what? In the IT domain, he beat us all as though we were a
>> bunch of click-and-drool folks. The guy /understood/ computing.
>
> My first few weeks at Uni, we didn't even go anywhere *near* the
> computer lab. We were taught to understand the *principles* of
> "problem solving" - i.e. breaking down tasks into abstractions, using
> syntax diagrams and plain English. Once we *did* get to the
> "programming" stage, we were taught to have the code completely
> *ready* and semantically proof-read *before* we even got to the lab
Must have been a long time ago. In this age of advanced IDEs that would
be madness.
Of course, have an overall functional spec done.
> and started typing it in. Our reports invariably had to include a
> plain English breakdown of what the code did ... and the proof and
> deductive reasoning of *why* it worked.
Nothing wrong with that. But this is college : the professors hardly
ever read the code. Something like Rational Rose can automatically parse
code and provide a top down view of the overall structure - that
combined with suitable comments is generally enough. Over commenting a
piece of SW (unless it has an API that needs to be made externally
available) is not necessarily a good thing and can lead to a lot of
discrepancies between fact and fiction : I find this particularly true
in the OSS community where the documentation for a project is frequently
pisspoor to say the least - there are no paid documentation departments.
>
> There's too many developers nowadays who just fire up an IDE and start
> pulling libraries and creating widgets ... before they even have the
> slightest *clue* what it is they're trying to create. This might
> account for why so much software is riddled with bugs ... on *all*
> platforms. However, the prevalence of "gooey" addiction on Windows
> might account for the apparently disproportionately *high* bug count
> on that platform.
Ditto for Linux : the GUI apps are very buggy - far more so than the
commercial offerings IMO.
>
> --
> K.
> http://slated.org - Slated, Rated & Blogged
> This message has not been photoshopped in any way.
>
> Fedora Core release 5 (Bordeaux) on sky, running kernel 2.6.16-1.2133_FC5
> 13:24:25 up 57 days, 13:41, 2 users, load average: 0.35, 0.53, 0.61
--
|
|
