Re: [boost] compile time asserts [preconditions]

From: John Maddock (John_Maddock_at_[hidden])
Date: 2000-08-01 05:46:34

Beman,

>Nits:

>*  While there isn't a law that requires it, lots of people expect all C++

>source files to begin with a comment line which simply says what's in the 
>file.

OK.

>*  I like a "Revision History" comments.  True, CVS supplies history, but 
>for casual reading a simple Revision History helps.  Or is that just 
>personal taste on my part?  I don't think we ever discussed it.

True it just hasn't been revised yet ;-)
OK, I'll add an "initial revision" line.

>*  How about adding a disclaimer in the docs to the effect: "Boost members

spent considerable effort trying to invent a compile time assert that 
avoided macros, all to no avail.  The general conclusion was that the good 
of a compile time assert working at namespace, function, and class scope 
outweighed the ugliness of a macro."  Maybe add at the bottom so it doesn't

take away from the central description.<

Good point.

>*  I had reading comprehension trouble with this sentence:

To avoid this, if you use BOOST_PRECONDITION in a header at namespace 
scope, then ensure that the declarations are enclosed in their own unique 
namespace.

Possibly clearer:

To avoid this, if you use BOOST_PRECONDITION in a header at namespace 
scope, enclose the use in a (possibly nested) namespace unique to that 
header.<

Yep.

>Basically, I think we ought to settle the name issue, formally review
this, 
>and start using it.

Yes, there isn't a whole lot to review here so it shouldn't be onerous (I
hope!)

With regard to the name, I'm fully committed to sitting on the fence on
this one.

It's clear that precondition was a bad choice, and the concensus (or as
close as we're likely to get) appears to be for BOOST_STATIC_ASSERT so
let's go with that.  That leaves the name given to the class that generates
the error (the class name that ends up in the error message), I favour
something more verbose for this - how about
"COMPILE_TIME_ASSERTION_FAILED"?

Finally, what header do we want this in?  Its own header
(static_assert.hpp??), or in utility.hpp or what?  There isn't a whole lot
of code here BTW.

-John.