Re: [ire] "resend" attribute in "deposit" element.

[Francisco Obispo <[e-mail]>]

Hi Alexander,

Well, if we have a detached signature that can be used to verify the integrity of the file, why do we neep to record the 'retries'?

I support the idea of removing the 'resend' attribute from the deposit, and in fact, I would go further and remove it from the file name as well.

Francisco




On Mar 29, 2013, at 7:27 AM, Alexander Mayrhofer <[e-mail]> wrote:

> Hi,
> 
> We're in the late stages of implementing our escrow export script, and have discussed internally the purpose of the "resend" attribute in the "deposit" element itself. 
> 
> Reading through the definition, this attribute is not related to the *creation* of the escrow deposit, but rather to its *transport* (Our understanding is that the id attribute is more related to the creation of the file). 
> 
> More specifically, we don't see a reason why the "resend" attribute should actually be contained in the file itself, because failure to "send" the file to the Escrow provider is completely unrelated to the contents of the file. 
> 
> If a transport attempt fails, a registry should simply be able to rename the (successfully created, validated) escrow file, and just retry the *transport*, rather than the *creation*. However, since the "resend" attribute is included in the (compressed/encrypted/signed) file itself (in addition to the file name!), a failed *transport* attempt (eg. a interrupted upload) requires registries to essentially restart the generation:  modify the original escrow dump, and repeat the whole split/compress/encrypt/sign steps. 
> 
> We hence believe that the "resend" attribute should be removed from the definition of the "deposit" element. It is not only redundant (since it's also included in the file name itself), but its inclusion in the escrow content itself is actually misleading, and puts extra burden on the escrow process. *If* ICANN's intent with the "resend" attribute was to reflect whether a file has been re-*generated* (and not just re-*sent*), we propose to change the current structure as follows:
> 
> - Keep the "resend" part of the file names, and provide clear instructions that this number is to be increased when *transport* of the deposit failed, and is repeated with the identical file contents (retransmission).
> - Remove the "resend" attribute from the "deposit" element.
> - (Optionally) add a "revision" attribute to the "deposit" element, and provide clear instructions that the number contained in that attribute is to be increased when the *creation* of a deposit with an identical "id" is performed (re-generation).
> 
> Feedback appreciated
> 
> Alex
> 
> 
> _______________________________________________
> ire mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/ire

Francisco Obispo 
Director of Applications and Services - ISC
email: [e-mail]
Phone: [phone] || INOC-DBA *3557* NOC
PGP KeyID = B38DB1BE