RE: I-D ACTION:draft-ietf-secsh-scp-sftp-ssh-uri-03.txt

["Salowey, Joe" <jsalowey%cisco.com@localhost>]

> There is an ambiguity here over whether the prohibition is of 
> overriding any cached host key for this session, or merely of 
> permanently replacing the cached host key.
> 
> Assuming both should be disallowed, "overwrite" should be 
> changed to "override" throughout.
> 

[Joe] Yes. The intent is not to automatically "override".  I think this
change should be made 


> | 3.2  SFTP Parameters
>   [...]
> |    typecode
> | 
> |       The typecode identifies the type of file which 
> determines how it
> |       will be treated.  [...]
> 
> No default behaviour is specified for the case where the 
> typecode is not present. This could lead to interoperability 
> problems similar to those which FTP has, where some clients 
> default to ASCII mode, and others to binary.
> 
> I think that "i" (no conversion) is an appropriate default in 
> the absence of further information; it's supported by all 
> filexfer implementations, and in any case is the "safe" thing 
> to do if nothing is known about the file.
> 
> I hesitate to mandate this as a default, however; 
> implementations should probably be free to use the 
> "text-hint" attribute to determine an appropriate transfer 
> mode for the file, and also to open directories in "d" mode.
> 
> How about the following text (new paragraph after the one above)?:
> 
>       In the absence of this parameter, files SHOULD be transmitted as
>       if "i" had been specified.  However, information from the server
>       (such as the "text-hint" attribute) MAY be used to select an
>       alternative mode.
>

[Joe] Thanks for the text, I think it looks good.