Re: No Go To's Forever!

[Robert Dewar <robert_dewar@my-deja.com>]

In article <8bd9ve$r1c$1@nnrp1.deja.com>,
  Ted Dennison <dennison@telepath.com> wrote:
> A tool? How would such a tool be written? It looks to me like
> you'd need
> an Ada parser to hook it to. Not everyone has the ability or
> permission
> to hack stuff like this into their compiler. Perhaps ASIS
> could help,
> but not every compiler supports ASIS.

Most compilers do support ASIS, and indeed this is a good
way or providing such checking tools.

This particular check of course is trivial to program, you
do not need ASIS

> And even if you did have the tool, how are you going to force
> maintainers to use it? Perhaps its a bit easier for you to
> make edicts like that at ACT, but I often write code that is
> owned by the government and will be maintained by whoever they
> decide to hire to do the job.

Maintainers who are under no constraints can do anything they
like, including replacing all your if statements with gotos.
Every significant body of code should be accompanied by a
rigorous set of coding standards, and competent maintenance
should maintain the same set of standards.

Yes, Ted, I know you have often complained of living in an
environment with limited ability compilers, lack of control
over procedures etc, but you should not assume that every
environment is like this :-)

> Really this is no more of a solution that "lint" is a solution
> to C's weak type checking.

You definitely need tools to enforce coding standards. There
are third party tools, e.g. the Xinotech Composer (hope I
am spelling name right) that can do a good job here.

As you know GNAT itself includes some extensive style checks
corresponding to a significant part of our internal coding
standards. Of course this does not include restrictions on
gotos, since we have no codified restrictions (it is merely
a feature in the language, that, like all other features,
should be used when and only when it is appropriate to do
so). Again you may complain that you work in an environment
where programmers cannot make this kind of decision correctly,
but that is not the universal situation.

Actually at ACT, we have a systematic procedure where all
changes made by one person are reviewed by another person,
for both logic and stylistic issues.

I consider such a procedure to be essential in any environment,
and it is even more important in a maintenance situation.


Sent via Deja.com http://www.deja.com/
Before you buy.