[Info-vax] Whither VMS?

[Richard B. Gilbert]

Bill Gunshannon wrote:
> In article <78Czm.16533$4Y3.13328 at newsfe25.ams2>,
> 	ChrisQ <meru at devnull.com> writes:
>> Richard B. Gilbert wrote:
>>
>>> The compiler is pretty good at spotting syntax errors.  Logical errors 
>>> or multiplying when you should have added are not something a compiler 
>>> can recognize as an error.
>>>
>>> Testing must find the logical errors.
>> The problem is the process. The number of programmers i've seen who, 
>> given a problem to solve. sit down and start editing code. Even for 
>> trivial programs, you should always design a solution on paper first, 
>> then the code just falls into place and often works first time. All the 
>> compiler should find should be missing semicolons etc.
>>
>> Think first, code later :-)...
>  
> Which brings us back to my other rant. The sorry state of what people
> today try to call Software Engineering (yes, I watch our SE Grad Students
> run straight from the classroom to the keyboard when given a programming
> assignment!)  maybe the main reason I have never had the problems with
> C that others are constantly lamenting here is because the development
> model I was taught in the very beginning of my career (which pre-dates
> the term Software Engineering by more than a decade) was just plain better.
> 
> bill
> 

Programming at the keyboard is not WRONG!  It is a poor way to design a 
large program but many problems can be solved with fewer than a hundred 
lines of code.  If you know the algorithm you can just sit down at the 
keyboard and code it or, better yet, pull something out of your source 
library and compile it.

The larger a software project is, the more it needs formal 
specifications, a design, etc.  If two or more programmers are going to 
be working on it the formal specifications and design are just about 
mandatory!  In this case Fred Brooks' "The Mythical Man Month" is 
mandatory reading!