Boost logo

Boost :

From: Pavol Droba (droba_at_[hidden])
Date: 2004-07-30 15:57:34


On Fri, Jul 30, 2004 at 02:07:05PM -0600, David Abrahams wrote:
> Pavol Droba <droba_at_[hidden]> writes:
>
> > On Fri, Jul 30, 2004 at 09:19:47AM -0700, David Abrahams wrote:
> >> Pavol Droba <droba_at_[hidden]> writes:
> >>
> >> > So will it be ok if I copy your definition from here:
> >> > <http://article.gmane.org/gmane.comp.lib.boost.devel/105915>
> >>
> >> You may, but I would prefer to update
> >> http://www.boost.org/more/error_handling.html and for you to post a
> >> link to that material.
> >
> > Don't you mean
> > <http://www.boost.org/more/generic_exception_safety.html> ?
>
> No I do not.
>

Ok, but I don't get it. The article you are refering is gereneraly
describing how to desing exceptions classe and how to throw an exception.

There is nothing even a marginaly close to an exception guarantie specification.
I assume, that this is what you want to add here.

But don't you think, that this is little bit unrelated to current problem?
AFAIK, we are trying to define a specification for a library. This specification
is primary oriented towards a user of the library, while this article is a collection
of good principles for a library authors.

In my understanding we need a proper definition of the exception safety concepts
what can be later used to label operations in the library.
 From the discussion it seems, that conditional concepts should be defined,
or the current concepts should be made conditional by default.

So I think, that we need a shorter version of generic-exception-safety document rather
then hide this information inside an unrelated doc.

I might have overlooked something. So please explain your rationale.

Regards,

Pavol


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk