Le lun. 1 sept. 2014 à 20:13, Andrea Faulds <a...@ajf.me> a écrit :
> > On 1 Sep 2014, at 17:29, Chris Wright <c...@daverandom.com> wrote: > > > It's also worth noting that the "return NULL on zpp failure" > > convention is not followed to the letter, I have seen places that > > RETURN_FALSE - I can't remember exactly where I have seen this but I > > will dig a ref out if anyone wants it. > > Good point. This is also, it has to be said, quite rarely documented. I’m > not sure we even document zip’s behaviour anywhere in the manual, which > means users don’t know what to expect. > > >> To help BC, we could even make it do the old thing if @ is being used > to silence it. > > > > Don't like this > > I’m not sure I do either, it’s merely a suggestion if it would help BC, > but it probably not a good idea, as @ should only suppress errors, not > change behaviour. > > > I'm roughly -0.5 on this proposal overall. I'd like this to be easily > > handleable in userland, but I don't like E_RECOVERABLE_ERROR (in > > general, not just here). I'd very much support this if it used > > exceptions, though. > > Perhaps this could be postponed until Nikita Popov can revive his > Exceptions in the Engine RFC. If that were to pass, I could make an RFC for > this. > -- > Andrea Faulds > http://ajf.me/ I don't have an opinion yet as I have to think a bit more about it. I have the impression that giving the user to bail out on specific error levels would be less intrusive as he would be able to opt-in for it. See the "Pedantic errors (Was: Re: [PHP-DEV] [RFC][VOTE] Improve array to string conversion)" thread. I would be glad to gather some opinions to see if it is worth investigating. Cheers, Patrick