David Binger wrote:
> On Aug 11, 2005, at 5:58 PM, mso@oz.net wrote:
>
>>> Should Quixote's publisher give the same result
>>> when PATH_INFO is 'foo/bar' that it gives
>>> when PATH_INFO is '/foo/bar' ?
>>> I don't know what to think about this, but it does seem good to
>>> do something other than trip on the assert.
>>>
>>
>> Well, what else could "foo/bar" mean?  It's really the fault of the
>> server, but we might as well make Quixote flexible for misplaced
>> slashes,
>> especially when the specs are not as clear as they should be.
>>
>
> I *suppose* that whenever the PATH_INFO is not a string
> that starts with '/', you could return a permanent
> redirect to SCRIPT_NAME + '/' + PATH_INFO
> This would cover the empty case the same way as this odd case.

Do we even need to code for that?  The observed problem is PATH_INFO="".
Are there any cases where PATH_INFO="foo/bar" has occurred?  Perhaps we
should raise an *informative* exception for that and wait for the
complaints to come in.  When we have a specific server situation to study,
we'll know better what to do.

--
-- Mike Orr