Re: [PEPr] -1 for RFC::Error Handling Guidelines for

Justin Patrin wrote:

On Mon, 30 Aug 2004 18:00:54 +0100, Sergio Carvalho <sergio.carvalho@portugalmail.com> wrote:

It's a moot discussion. For good and evil, the RFC has been approved, following the standard procedure for PEAR. This is the basis for work we have now, so let's build on it and correct where needed. There's much work to be done defining warnings, and this should start with a package proposal -- unlike exceptions, warnings will require some infrastructure, and naturally there are lots of design possibilities.

[snip] A package proposal?? ErrorStack is already done and in PEAR. All of the comments I've seen on warnings advocate the stack method, or at least multiple warnings "at once", so the stack would work for it.

I think that everyone (esp Greg) has acknowledged that to be designed for warnings, PEAR_ErrorStack would need some [perhaps minor] modifications. I remember, in particular, the suggestion that the stack approach be able to use Exception objects to provide for easy promotion / demotion of warnings. There were some other proposed changes that I'm too lazy to search on atm. Plus the name ErrorStack would be a little misleading if the system was designed for warnings. Anyway, Sergio is right in saying that this should be a separate package / proposal, although I think you are right that most everyone agrees a stack-based approach will work exceptionally well for warnings. Cheers, Hans