- From: Marcel Fröhlich <marcel.frohlich@gmail.com>
- Date: Thu, 27 Apr 2017 17:49:27 +0200
- To: Stephane Fellah <fellahst@gmail.com>
- Cc: Keith Alexander <keithalexander@keithalexander.co.uk>, Sarven Capadisli <info@csarven.ca>, Linking Open Data <public-lod@w3.org>
- Message-ID: <CAHKA4LzWeVjxxwmuSOS=RSyE=-X0-GmYD-jPvwLusS3tj60nXA@mail.gmail.com>
Read first, answer later ... I do understand the difference between naming and addressing. And yes, you can have a "http://" substring as part of a name without creating an immediate problem. But then why not keeping it separate, because sometimes you want to change the access method, and looking at the http vs https discussion, you may end up with problems later, e.g. because of some meta characters. Or is it straight forward to access the http-URI using https protocol without applying any additional tricks? Regards, Marcel 2017-04-27 17:39 GMT+02:00 Marcel Fröhlich <marcel.frohlich@gmail.com>: > Stephane, > > I did not advocate anything else. > My point is that naming data points and providing an addressable linked > data handle is usually not distinguished. > I argued that in some more critical use cases (e.g. regulated industries) > it is worth while to separate this clearly. > > Regards, Marcel > > 2017-04-27 16:57 GMT+02:00 Stephane Fellah <fellahst@gmail.com>: > >> Please keep in mind that in the context of Linked Data, the use of HTTP >> scheme in URI is just fine and actually recommended, as URI should be >> considered only as globally unique names identifying a concept. Naming and >> addressing are two different things. Here the excellent blog for Norman >> Walsh that explains the difference. https://norman.wal >> sh.name/2006/07/25/namesAndAddresses. >> >> Regards >> Stephane Fellah >> >> >> >> >> On Thu, Apr 27, 2017 at 10:35 AM, Marcel Fröhlich < >> marcel.frohlich@gmail.com> wrote: >> >>> Instead of policing a refined design could help (for use cases that are >>> worth the additional effort) >>> >>> - As unique identifiers URNs may be cooler than URIs because they don't >>> mix up naming and access protocol. >>> Expect more protocol changes/variety to come than just a switch from >>> HTTP to HTTPS, even if it takes some time. >>> >>> - Vanilla DNS is a mediocre base for URI namespaces, because DNS >>> governance allows re-assignment of domains. >>> We need top-level domains that assign domain names only once. >>> >>> Marcel >>> >>> 2017-04-27 15:43 GMT+02:00 Keith Alexander < >>> keithalexander@keithalexander.co.uk>: >>> >>>> When a naming system requires names to be maintained by an owner at >>>> cost (domain renewals, server provision, etc), it's very likely that names >>>> are going to change and fall out of use a lot... >>>> >>>> "Policing" (via pedantic-web or otherwise) may be useful for accidental >>>> name changes, but otherwise hints at a conflict of interests between >>>> publisher and consumer (ie, the consumer wants the old name, the publisher >>>> wants the new one or wants to stop supporting the name altogether). >>>> >>>> On Thu, Apr 27, 2017 at 1:24 PM, Sarven Capadisli <info@csarven.ca> >>>> wrote: >>>> >>>>> Are "cool URIs don't change" for life? >>>>> >>>>> Would the policing of this fall under the jurisdiction of pedantic-web? >>>>> >>>>> Discuss. >>>>> >>>>> Aside: Please help me decide on this burning issue that I've been >>>>> putting off: https://twitter.com/csarven/status/857569335908454401 >>>>> >>>>> -Sarven >>>>> http://csarven.ca/#i >>>>> >>>>> >>>> >>> >> >
Received on Thursday, 27 April 2017 15:50:02 UTC