Re: comments regarding versioning

On Tue, Jun 30, 2015 at 8:36 PM, Erik Wilde <dret@berkeley.edu> wrote:
> hello herbert.
>
> On 2015-06-30 09:50, Herbert Van de Sompel wrote:
>>>
>>> if you think they're similar but not actually the same, what would you
>>> say
>>> the difference is? i am still struggling to see it, and if they do remain
>>> separate, it might be helpful to better explain how they are different.
>>
>> Well, the difference is in the nature of the content, by which I mean
>> that I have a hard time thinking about a vocabulary as a dataset. But
>> other than that, when it comes to versioning, both datasets and
>> vocabularies are just resources that can be versioned, and to which
>> Memento patterns for version disclosure and version access can be
>> uniformly applied.
>
>
> that's funny, i have a hard time thinking about a vocabulary *not* being a
> dataset. metadata is data, too, it's all a question of perspective. so
> there's no need to make any special rules for something that in some context
> happens to be metadata. many publishers may choose to not expose
> vocabularies in the same way as data (because for example they may not have
> control over them), and i think that makes a lot of sense, but if they do
> choose to publish them as data, normal rules apply.
>

I think we agree regarding the essence: the same versioning patterns
can be applied to data, vocabs. I remember you making a comment about
API versioning. API descriptions, e.g. using RESTdesc
(http://restdesc.org/about/descriptions), could be versioned using the
same pattern.

>> Overall, I think I should be happy with the fact that Memento is being
>> mentioned with regard to versioning. On the other hand, I am not sure
>> why alternatives such as:
>> - the use of an "API for version access" mentioned Best Practice 8
>> - the use of an "API for access to version history" mentioned in Best
>> Practice 9
>> should be promoted, understanding that there is an RFC (7089) (which
>> was 4 years in the making) that provides - IMO - all the required
>> capabilities and is fully aligned with REST, follow-your-nose. I guess
>> this question is related to what the ultimate goal of this document
>> is.
>
>
> memento *is* an API, right? so some may choose memento, others may
> choose
> other things. BP documents should stay away from starting to recommend
> specific technologies, that usually leads to a lot of fighting by technology
> owners/creators to get their techs in. so i think you're in good shape
> there!
>

agreed. your comment relates to my question about the goal of the doc.
and I tend to agree that a BP doc should probably not recommend
specific technologies. at the same time, they need to give
sufficient/meaningful guidance.

>> Maybe Memento is not understood well enough yet. So, maybe it is not
>> clear to many how it addresses all these versioning requirements. I
>> would be most happy to provide additional information, if that would
>> be considered helpful.
>
>
> like i said, i think you're in good shape there. memento is an RFC which is
> great, and if people try to follow BP, they should be smart enough to
> discover it as one way to solve some of the problems mentioned in the
> document.
>

OK

> what concerns me more right now is the lack of a webby perspective.
> curiously, the current WD does not even mention the terms "hypertext" or
> "hypermedia", which i find rather odd for a document that tries to define
> best practices for publishing anything on the web. the current spec is
> really more *what* to put *on* the web, not so much about *how* to make
> something to be part *of* the web.
>
> to some extent, that's what linked data is doing, but then again, that's
> something that champions a specific set of techs, which like i said probably
> shouldn't be part of any BP. this is the reason why i started this little
> project here, which i should probably switch from being a repo to being an
> actual page/site:
>
> https://github.com/dret/webdata
>

I think this is actually very nice, as you indicate, in the way it is
technology-neutral.

(looks like it's only us talking currently)

Cheers

Herbert

> it's basically talking about how to be nicely webby data, without making any
> assumptions which tech you should choose.
>
> and since hypermedia is such a central part of being webby, another little
> project is here, which tries to classify the ways in which data can be
> hypermedia-ish (ie, webby), and then lists some formats. this one needs a
> lot work (and probably more structure), but i think it can help people a lot
> to better understand how to make data properly webby, without having to
> reinvent the wheel:
>
> https://github.com/dret/hyperpedia
>
> i hope i will get some feedback about the issues i raised, in particular the
> curious absence of webbiness in the current WD. i think it would be great to
> better explain to people that making data webby is a better BP than just how
> to make your data available for download.
>
> cheers,
>
> dret.
>
> --
> erik wilde | mailto:dret@berkeley.edu  -  tel:+1-510-2061079 |
>            | UC Berkeley  -  School of Information (ISchool) |
>            | http://dret.net/netdret http://twitter.com/dret |



-- 
Herbert Van de Sompel
Digital Library Research & Prototyping
Los Alamos National Laboratory, Research Library
http://public.lanl.gov/herbertv/

==

Received on Wednesday, 1 July 2015 17:30:54 UTC