RE: [boost] RE: Serialization Library Review - archive exception

From: Jeff Garland (jeff_at_[hidden])
Date: 2002-11-17 13:14:33

> 1) I made orginally made archive_exception the simplest possible, it wasn't derived
> from std::exception and and contain an enum of every exception type.  It suited
> my needs and I didn't feel that std::exception added anything.

Not for you, but for clients of the library that want a reasonable exception
handler of last resort and don't want to clutter code with all sorts of
exception types it is important. 
 
> 2) well lots of complaints, some people didn't know about catch(...) syntax
> and thought that it had to be derived from std::archive to have a catch all.
> Others thought the embedded string was important.

I know about the syntax, but I really can't do anything with in the ... clause
because I can't know anything about the exception.  Thuse, it is totally 
useless in practice.

> 3) I agreed with none of these things but what the hell, its easier to
> accomodate than make a big deal on this insignificant point.  Its obvious
> that in retrospect I was wrong.

No you were right, I would be taking issue if you hadn't done this...

> In general, libary code should make no presumptions as to the language
> of the user.  That means not embedded messages. 

Yes, we need to provide locale indexed message strings. No debate on that. 
Sounds like another requirement for boost::exception.

> If you want a key in to a message table, use archive_exception::exception_code
> that's what my intention was.
> 
> In my view std::exception is a mistake and should be removed from the standard.

In my view it needs to be enhanced to support developers needs. 

Jeff