mugle-dev team mailing list archive

[Matt Giuca <matt.giuca@xxxxxxxxx>]

> when recieving "#!/~somestring" as a url what is the best method for
> determining whether "somestring" is the name of a User or a Devteam?
> The only things i can come up with are:
>  1. Query the database to see if a user with that name exists. If it
> doesnt, try devTeams.
>  2. Hava different symbol for each
>

Right ... they are sharing a namespace so by definition, we can't be sure
from the name which one we are dealing with. I think option #1 is fine,
since as I said yesterday, we don't expect any users to ever conflict with a
devteam. If anyone really wants to veto this, we can go with a different
symbol. We could potentially use the hyphen (/-myteam) though that looks a
bit weird.

For what it's worth, there is a formal reason to use tilde. RFC
3986<http://tools.ietf.org/html/rfc3986>(URI syntax) defines
"unreserved characters" (section 2.3) as:

unreserved = ALPHA / DIGIT / "-" / "." / "_" / "~"

This means that the only non alphanumeric characters which may be used in
URLs (including all possible Unicode characters) are the hyphen (-), the
period (.), the underscore (_) and the tilde (~). Everything else is
reserved for use in URI syntax, or non-ASCII which is not allowed. All other
characters must be percent-escaped. I suppose, though, that we are already
breaking this rule with the + character. As I see it, all the other
characters look very weird as part of generic syntax in front of a name.

An alternative approach is that rather than unify the user and devteam
namespaces, we unify the devteam and game namespaces. After all, they will
only share when we promote a game, so we can change the name of a game we
promote if it conflicts with a devteam, and we can avoid naming devteams
after games. That way, there will be no funny characters in the URL of a
promoted or non-promoted game. It will be:

/myteam/mygame/

I'm for that approach.

By the way, I think I figured out the significance of the "!" which is not
really necessary except by convention. Apparently, the Google web crawler
recognises URLs with the "#!" in them as distinct pages, allowing it to
crawl web pages that are rendered entirely using Ajax. It can't do that for
all # URLs, because that would include any page with anchors (such as all
section headings on Wikipedia).