[bad_user]

> Because everything from the web is a string.

I would start by saying that everything, regardless of domain, is just a stream of bits. Which is completely useless, just like your assertion.

And I know what you meant, but you're also wrong. A JSON object is only a string before being interpreted. An x-www-form-urlencoded is actually a map in which values can be arrays instead of primitives. Such forms often correspond to domain models with a clear definition.

There's also no such thing as "obviously bad data". All data is good in the proper context, therefore automatic conversions that try to make this distinction do not make sense. I don't necessarily know how PHP behaves, but in another popular language there's a world of difference between "077" and "77". There's also a world of difference between integer, floating point and fixed point and the details are never irrelevant.

[ars]

> therefore automatic conversions that try to make this distinction do not make sense.

You have to convert it somewhere. I don't see how the caller converting it is any better than the recipient doing it. Having the caller do it seems quite pointless when the recipient is anyway doing it.

Your answer about how everything is bits was quite useless since you completely missed the point. Your input is a string, you have to convert it someplace. Weak mode has the recipient do it. Strict mode you have to do it yourself, and then the recipient double checks.

I see no value in the second option - the actual conversion in both cases is identical.

But Danack disagrees, so I asked him to explain. Your answer was not helpful at all.

[bad_user]

@ars, you completely missed my point. Lets go over it again. Tell me how should the following things convert:

     77
     077
     77.0

[ars]

I GOT your point. I don't care about your point because it is not the question I am asking.

Why are you answering something I did not ask?

I am asking why does Danack prefer strict mode. There is absolutely nothing in my question that cares about the specific details how you convert bits to types, other than that you do.

My question is entirely about WHO does the conversion. NOT about the conversion itself.

(Oh, and the thing about bad-data has a defined meaning that went over your head because you are not familiar with the debate here. In this context bad-data means data loss on conversion. So "1" to 1 is fine, but "1 a" to 1 is not.)

[bad_user]

The conversion itself is very relevant because you cannot establish a default conversion that should happen, therefore the conversions should be explicit, answering the WHO. This is why I asked you about what should the conversion produce in those examples.

And also in the conversion from "1.1" to 1.1 there is loss of information, because the two representations are not isomorphic. Care to guess why?

[marrs]

This is the point that ars is making. He is saying that it has to be a runtime check and/or conversion because it is coming over the wire, as a string, at runtime.

So the question he is asking is, if you have to do the check at runtime anyway, what is the benefit of the type hinting? Isn't it just belt and braces?

It seems like a perfectly legitimate question to me.

And by the way, to those downvoters who don't seem to be able to tell the difference between a comment you disagree with and spam, can you please contribute to the conversation by hitting the reply button or alternatively get lost? Only you're ruining it for the rest of us. Thanks

[ircmaxell]

> Having the caller do it seems quite pointless when the recipient is anyway doing it.

Actually, I disagree. The caller is the only one who has semantic information about what the variable (and hence its value) means. All the callee (recipient) can do is blind cast it. The caller on the other hand can interpret it because it knows the meaning (talking about the developer, not the engine).