Print

Print


I tried to answer an off-list message, but it bounced.

> ----- Transcript of session follows -----
> 421 prime.imagin.net.: Connection refused by prime.imagin.net.

   ----- Unsent message follows -----
Date: Sun, 29 Nov 1998 22:26:46 -0500
From: "W. Wesley Groleau x4923" <[log in to unmask]>
To: [log in to unmask], [log in to unmask]
Subject: Re: object.method notation

> I don't want some idiot tool telling me, over and over, that a given
> unit fails some design-level semantic check -- especially since I may
> have a whole raft of checkable paradigms, and they may be inconsistent
> with each other.  Consider how unpopular "lint" is with C developers.

I thought 'lint' was unpopular for the same reason Ada is unpopular.  :-)

> > I still would proceed slowly on "standardizing" new pragmas for stylistic
> > purposes.  As I said, if we introduce several pragmas to support one
> > stylistic paradigm, how many other styles might also want pragmas?
>
> I'm easy with comment flags if those would get through ASIS.  Does
> that address your concern?
>
> I personally would hope that people would want lots and lots of flags
> to make the compiler help them ensure they did things right as they
> implemented their design-level decisions into code.

Whether comments or pragmas, my point was to avoid having lots and lots of
flags imposed on me by the compiler vendor because another customer wanted
them.  Multiply that by two or three customers, and the flags "I" want are
hard to find--or maybe even conflicting with those already defined.

But if you write (or buy) an ASIS tool that implements your methods,
independently of the compiler, you are free to give it or sell it to
anybody else who wants it.  If that turns out to be everybody, so be it.
But if not, it doesn't get in the way of those that want to do something
else.

(Plus, it doesn't tie you to one compiler.)


----- End Included Message -----