[dev] Re: ignoring the url generation bug in imp

[Chuck Hagenbuch]

Quoting Marc Lehmann <pcg@goof.com>:

> Almost universally always I would think, as the browser constructs GET
> requests most of the time (in forms), unless you hardcode get-requests
> insise urls (as horde does most of the time ;)

Hmmm, okay, true - for some reason I was forgetting about GET from forms. ;)
Question, though: how many browsers actually use ";" as the arg seperator?

> No. The format of GET requests is standardized by the HTTP protocol and
> every server MUST implement both "&" and ";" forms. If php fails to do that it
> doesn't speak http.

Interesting. I guess this makes sense; for some reason I was thinking that it'd
be impossible to handle either...

You might want to bring this up on php-dev.

> This is true. However, the support for ";" and "&" is there since http/0.9
> and earlier and cgi scripts can (and did) use whichever seperator they
> prefered. HTML 4 just now *recommends* this, but this does not mean that
> this isn't supported since the old ages of http. As such, I am puzzled as
> why this has to be configured on php.

I'm coming around to your point of view. Good question.

> My point is that *maybe* (and it seems it isn't) it is easier to just use
> ";" as the seperator when *you* create the URLs yourself, as the server
> (or php in this case) must parse both forms anyway and ";" needs not to be
> escaped.

I'm convinced; if we could get a change into php to support either, then it
would make sense to use ";".

-chuck

--
Charles Hagenbuch, <chuck@horde.org>
"If you can't stand the heat, get out of the chicken!" - Baby Blues