Re: [Tools-discuss] Unversioned draft links (was Re: slow tools reaction)

[Robert Sparks <[e-mail]>]

On 9/19/21 8:32 PM, Martin Thomson wrote:
> Thanks Robert,
>
> (Changing the subject as we are now way off the original subject.)
>
> Is there a resource that refers to the latest text rendering of a draft regardless of version?

No, there is not a good single resource for this, but I am in the 
process of making one.

See the thread at 
<https://mailarchive.ietf.org/arch/browse/tools-discuss/?q=Getting%20the%20%22current%22%20version%20of%20things>

You are, I think, stating a preference for {versionlessname}.{extension} 
(and inferring format from extension) in whatever URL plan we end up 
with in the long run vs {versionlessname}/{format}?

Similarly, {versionlessname}-{version}.{extension} vs 
{versionlessname}/{version}/{format}

As a brief diversion, while thinking about this, consider the foot-wound 
we carry around that is heuristic extraction of version given a string 
that looks like aaaaaa-nn.txt.

Things like this 
<https://datatracker.ietf.org/doc/draft-ietf-16ng-ip-over-ethernet-over-802-dot-16> 
make for code that has to guess at what a request for 
"draft-ietf-16-ng-ip-over-ethernet-over-802-dot-16.txt" means. We 
heuristically solve this by looking for -802-dot version 16 first and if 
we don't find it, we look for -802-dot-16 version latest. There's a 
chance for harmful masking of a name here - fortunately we haven't seen 
it exercised in practice.

And then consider what horror we inflicted on ourselves with names like 
review-ietf-taps-interface-13-artart-early-sparks-2021-09-17-00 and 
charter-ietf-quic-02-06.

RjS

>   
>
> The only unversioned links I'm aware of point to the latest HTMLized version of a draft and its metadata page.  I'm in the process of converting the links I generate to use this version of rfcdiff and some of the links I used to provide I find are now broken.
>
> For example, this is a link that used to be useful.  Note the absence of a -06 in url1:
> https://tools.ietf.org/rfcdiff?url1=https://tools.ietf.org/id/draft-ietf-httpbis-message-signatures.txt&url2=https://httpwg.github.io/http-extensions/draft-ietf-httpbis-message-signatures.txt
>
> I can now fix the first part of that, but url1 remains a problem.
>
> (Maybe the fact that no one has complained about these links being broken is a sign that maybe there isn't enough use of this feature to continue supporting it, but I thought that I would at least try.)
>
> On Fri, Sep 17, 2021, at 12:44, Robert Sparks wrote:
>> Ah - I see at least one place where it leaks (in the instructions at the
>> bottom of the default form - I will be sure that is addressed. Thanks
>> for calling attention to it - if you see it anywhere else, please let me
>> know.