On Sun, 15 Feb 2015, Nicholas Boel wrote to Wilfred van Velzen:

 WV> Ok, my opinion:

 WV> The processor can't know it's a typo. It can only see that the file to
 WV> be replaced doesn't exist on the receiving system. That's a legitimate
 WV> (local) situation and not an error in itself.

 NB> The I can only be led to think that the processor that actually
 NB> stopped this assumed the comma was a bad filename. With that, I can
 NB> pass along the information to the other developer to allow for a
 NB> comma in the filename, since it's allowed on both Linux and
 NB> Windows.

just remember that it breaks on DOS and when passing to DOS machines,
certain additional steps must be taken... i remember when there was work
being done for transferring LFNs where the system had to choose which name
format to send to which remote system... 8.3 for those that could not
handle LFNs and LFNs for those that could... i'd have to dig for the
specifics as it was many years back and i don't recall if it went
mainstream or not but there were many systems operating with that code...
that's also one reason why most of the FDNs still use 8.3 names ;)

)\/(ark