From: Beman Dawes (beman_at_[hidden])
Date: 2000-03-18 17:26:14
At 10:26 PM 3/17/00 -0500, Dave Abrahams wrote:
>We eschew exception-specifications at boost. For the reasons why,
>[Sheesh, Beman, I guess this comes up often enough that we ought to
>rationale on the site. Egroups is a lovely host but their search
>leaves something to be desired. I can't find the original posting.]
>In brief, they provide no compile-time safety and lead to worse
>code on many (if not most) compilers.
Yes. I will add something to the library guidelines. If you would
like to expand the "In brief," above to a paragraph or two, I will
I am starting to experiment more with HTML documentation,
particularly supplying a simple statement like: "Boost guidelines are
to avoid exception-specifications" with a link to more details
I believe that not supplying rationale is a major defect in many
software projects. Lack of available rationale causes issues to
revisited endlessly, causes all sorts of maintenance bugs when a
programmer changes something without realizing it was done a certain
way for a purpose, and shortens the useful lifetime of code.
Rationale is fairly easy to provide at the time decisions are made,
but very hard to recover even a short time later.
I guess the docs for proposed boost libraries should be reviewed to
see that there is rationale for major design decisions.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk