We should consistently ensure bugtracker URLs are slash-terminated (or not)

What do you mean by "generating links manually" ?

Is the idea that a bugwatch URL will always have things added to the end of it? If so, let's make it always end in a slash.

As a note for the future: there's a URL class in the menus code that we can use more generally. It provides an API for doing things with URLs like getting a version of the URL that always ends in a slash.

Well, bugwatch.py:BugWatch does string manipulation to generate URLs to remote bugs. And yes, the bugwatch URL will have bits added to its end, so it makes sense to make it always end in a slash.

Would it be better to use this URL class now, then?

Taking into account all the comments, I think we should ensure that the URLs always end in a slash, and that this will be documented in the appropriate interfaces and doctests.

I'd prefer that they always *not* have a slash, for brevity and easier typing. Requiring a slash wouldn't have prevented bug 1479, and it may not even work for some bug trackers.

Actually, using urlparse.urljoin avoids the whole nonsense about validating URL termination -- enter it any way you like and it always works.

On Sat, Apr 08, 2006 at 03:51:27PM -0000, Christian Reis wrote:
> Actually, using urlparse.urljoin avoids the whole nonsense about
> validating URL termination -- enter it any way you like and it always
> works.

urljoin won't help here. It makes a huge difference if the URL ends
with a slash or not.

    >>> urlparse.urljoin('http://foo.com/bar', 'baz')
    'http://foo.com/baz'

    >>> urlparse.urljoin('http://foo.com/bar/', 'baz')
    'http://foo.com/bar/baz'

    status confirmed

Sorry -- yeah, you're right. Fixed to use helpers.urlappend, and included more specific tests now.