This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

Author r.david.murray
Recipients amrith, eric.smith, martin.panter, orsenthil, r.david.murray, serhiy.storchaka
Date 2016-12-01.04:40:55
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1480567256.24.0.451977773277.issue28841@psf.upfronthosting.co.za>
In-reply-to
Content
Yes, I'm afraid so.  But the rationale is presumably the same as it is in many RFCs: you should accept broken stuff if it can be interpreted unambiguously, but you should not *produce* the broken stuff.  So I'd say the RFC is correct as quoted.  It is then up to openstack whether or not it wants to accept such URLs in a given interface, and it if were me I'd base that decision on whether it was an 'internal' interface or an 'external' one.  But you can also argue that even an external interface could be strict, depending on the application domain of the interface.  urllib, on the other hand, needs to accept it, and we can't change it at this point for backward compatibility reasons if nothing else.

Based on the RFC, one could argue that urlunsplit should omit the colon.  That could break backward compatibility, too, though will a lot less likelyhood.  So we could still fix it in 3.7 if we decide we should.
History
Date User Action Args
2016-12-01 04:40:56r.david.murraysetrecipients: + r.david.murray, orsenthil, eric.smith, martin.panter, serhiy.storchaka, amrith
2016-12-01 04:40:56r.david.murraysetmessageid: <1480567256.24.0.451977773277.issue28841@psf.upfronthosting.co.za>
2016-12-01 04:40:56r.david.murraylinkissue28841 messages
2016-12-01 04:40:55r.david.murraycreate