Re: Approaching the First Draft [of a language spec]

> Well, I guess I see that a bit the other way around. I see the goal as a > document that can make the user community know what the language definition > is. The kind of document you are talking about, which is similar to say, > ANSI C/ANSI C++ specs, are specs that are useful for compiler developers. > > Today, quite often, people come out with questions of 'what happens in this > case'. For instance, someone asked me today on IRC whether !$foo checks if > $foo is false, or if $foo is set. Today, there's no place he can go to and > find an answer to his question. The way I was imagining the language spec, > I thought it would address exactly those needs (well, that was just a > simple example :). > > Syntactical rules can be useful to end users, and they're common in > programming language books. If we add a description of the semantics > surrounding these rules, we can end up with a document that would be useful > for the average PHP programmer. Well, so we almost agree here. Syntactical rules are useless without a context which defines their exact semantics (see Task 2 of my proposal). That is the format which is chosen by the C/C++ standards as well as the Python Reference. I disagree though that the immediate result of this process must be a document which can be understood by your average IRC user. We have the manual for that. - Sascha