First draft of specifications

Dear all,
I've extended Thomas' draft with my part on models. I took up Frederik's
suggestion to look at this as traits as this allowed me to circumvent
category issues and also explain some cases through combined traits. See
attached. Talk to you tomorrow :-)
Best, Tobias
-------- Original Message --------
Subject: [rda-collection-wg] First draft of specifications
From: ThomasZastrow
<***@***.***>
To: RDA Collections WG <***@***.***-groups.org>
Date: 02 May 2016, 15:40

... via GoTo meeting?

Dear colleagues,
I could not manage to sent this to the group together with
specifications document containing my change suggestions. The rda
mailing list seems to have problems. Therefore I try it now without that
document.

Hello Ulrich,
thank you for extending the traits and opening up various discussion
items. I agree that the current section in the document is probably hard
to understand without the discussion history. More explanation is
definitely needed.
I have some comments and questions on what you wrote:
1.) I suggest to define "Persistent Collection" in the document. Even I
still tend to mix up the persistency of object, identifier and
collection, even though your argumentation is clear on this. Defining
these may resolve the open discussion on this point.
2.) Regarding the "mutable" trait you added: I had this in as a trait
originally, but replaced it with extendable and shrinkable. What is the
mandatory property (you did not include a name)? I also now think that
exclusive membership should be under the "*" trait - even a collection
that is once created and never changed can follow or not follow this policy.
3.) I frowned when reading about your ideas on mutability of rule vs.
mutability of the resulting collection, but you are right. There are
significant differences. Perhaps my problem in understanding this
entirely is that the only solid example I currently have for a
rule-based collection is the "powercollection" you brought up earlier.
Can you invent more examples? Drilling down on these may help me resolve
the categories you described. Anyway, I agree that the second case is
probably the most common one/most practically relevant.
- One defining difference I see is that, assuming a rule is included as
part of the collection definition, it can be made read-only quite
easily, while a resulting collection's immutability is only
guaranteeable through a PID provider's (or similar) QA policies
(ignoring for now how these are going to be implemented - tough!).
- Also, a point to keep in mind in general: The more mutable a
collection is, the less sense it may make to put a PID on it. In the
end, we run (again) into the philosophical question of whether the PID
identifies a conceptual entity (persistent over time) whose concrete
manifestation is subject to various changes (in rule definition and
concrete membership). I am not sure how far we want to go down that
road; I currently see that most of our use cases would be fine with a
more static interpretation.
Perhaps the most important question: Even if I have a rule-based
collection whose rule may change, how will this work with "adding a
single arbitrary item to the collection"? That is the key reason why I
saw such combinations as invalid. Even with your 4 categories listed, I
still understand that a rule-based collection is only extendable by
changing its rule or re-applying it to a changed "base heap of objects",
but it would not be possible to extend it by adding a single item
(because: how would you redefine the rule for this?).
4.) Perhaps a solution to 3., we should define "rule-based collection".
My first shot is that I see the rule as part of the "construction
definition" of a collection, but I am not sure as changing such a rule
means changing the collection definition. Again, a point towards the
philosophical question already mentioned.
5.) The invalid trait combinations now in the document are by definition
impossible due to the semantics of the "extend" facility.
Regarding the diagrams: I included them for further reference, but they
definitely need to be revisited. I perceive the "streaming" idea as a
special application scenario that I may describe with traits at some point.
Best, Tobias
-------- Original Message --------
*Subject: *Re: [rda-collection-wg] First draft of specifications
*From: *Ulrich Schwardmann <***@***.***>
*To: *TobiasWeigel <***@***.***>, ThomasZastrow
<***@***.***>, RDA Collections WG
<***@***.***-groups.org>

Also, for the call today, if we want to discuss the API, the latest
version can be seen by launching
http://rdacollectionswg.github.io/apidocs/# and changing the path to
swagger file to
https://raw.githubusercontent.com/RDACollectionsWG/apidocs/operations/sw...
(to retrieve from the operations branch, not yet merged to master)
Best

.. also wondering: I'm currently using the link for the montly call, but
not sure if that is going to work as it says "waiting for the RDA
secretariat"...
-------- Original Message --------
*Subject: *Re: [rda-collection-wg] First draft of specifications
*From: *ThomasZastrow
<***@***.***>
*To: *TobiasWeigel <***@***.***>, RDA Collections WG
<***@***.***-groups.org>

We are having trouble with the RDA gotomeeting account. We will have to
use google hangouts for this call. Here's the link:
https://hangouts.google.com/call/ainf3nrrv5agpjf7liioqi33l4e

Sorry all, we're having trouble with the Goto meeting account. We have
to use hangouts
https://hangouts.google.com/call/ainf3nrrv5agpjf7liioqi33l4e

I also started that meeting, it says that it is re-occurring

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Same problem here...
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Same problem here...
Am 10.05.2016 um 15:06 schrieb TobiasWeigel:
> .. also wondering: I'm currently using the link for the montly
> call, but not sure if that is going to work as it says "waiting for
> the RDA secretariat"...
>
> -------- Original Message -------- *Subject: *Re:
> [rda-collection-wg] First draft of specifications *From:
> *ThomasZastrow
<***@***.***> *To: *TobiasWeigel
> <***@***.***>, RDA Collections WG
> <***@***.***-groups.org> *Date: *10 May 2016, 09:09
>> ... via GoTo meeting?
>>
>>
>> Am 09.05.2016 um 17:51 schrieb TobiasWeigel:
>>> Dear all,
>>>
>>> I've extended Thomas' draft with my part on models. I took up
>>> Frederik's suggestion to look at this as traits as this allowed
>>> me to circumvent category issues and also explain some cases
>>> through combined traits. See attached. Talk to you tomorrow
>>> :-)
>>>
>>> Best, Tobias
>>>
>>> -------- Original Message -------- Subject:
>>> [rda-collection-wg] First draft of specifications From:
>>> ThomasZastrow
<***@***.***> To: RDA Collections
>>> WG <***@***.***-groups.org> Date: 02 May 2016, 15:40
>>>
>>>> Dear all,
>>>>
>>>> So maybe its the best way to send you the first draft of a
>>>> specification document via good old email ;-)
>>>>
>>>> Best,
>>>>
>>>> Tom
>>>>
>>>>
>>>>
>>>> -- Full post:
>>>> https://rd-alliance.org/group/research-data-collections-wg/post/fir
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Same problem here...
Am 10.05.2016 um 15:06 schrieb TobiasWeigel:
> .. also wondering: I'm currently using the link for the montly
> call, but not sure if that is going to work as it says "waiting for
> the RDA secretariat"...
>
> -------- Original Message -------- *Subject: *Re:
> [rda-collection-wg] First draft of specifications *From:
> *ThomasZastrow
<***@***.***> *To: *TobiasWeigel
> <***@***.***>, RDA Collections WG
> <***@***.***-groups.org> *Date: *10 May 2016, 09:09
>> ... via GoTo meeting?
>>
>>
>> Am 09.05.2016 um 17:51 schrieb TobiasWeigel:
>>> Dear all,
>>>
>>> I've extended Thomas' draft with my part on models. I took up
>>> Frederik's suggestion to look at this as traits as this allowed
>>> me to circumvent category issues and also explain some cases
>>> through combined traits. See attached. Talk to you tomorrow
>>> :-)
>>>
>>> Best, Tobias
>>>
>>> -------- Original Message -------- Subject:
>>> [rda-collection-wg] First draft of specifications From:
>>> ThomasZastrow
<***@***.***> To: RDA Collections
>>> WG <***@***.***-groups.org> Date: 02 May 2016, 15:40
>>>
>>>> Dear all,
>>>>
>>>> So maybe its the best way to send you the first draft of a
>>>> specification document via good old email ;-)
>>>>
>>>> Best,
>>>>
>>>> Tom
>>>>
>>>>
>>>>
>>>> -- Full post:
>>>> https://rd-alliance.org/group/research-data-collections-wg/post/fir
st-draft-specifications
>>>>
>>>>
>>>>
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Same problem here...
Am 10.05.2016 um 15:06 schrieb TobiasWeigel:
> .. also wondering: I'm currently using the link for the montly
> call, but not sure if that is going to work as it says "waiting for
> the RDA secretariat"...
>
> -------- Original Message -------- *Subject: *Re:
> [rda-collection-wg] First draft of specifications *From:
> *ThomasZastrow
<***@***.***> *To: *TobiasWeigel
> <***@***.***>, RDA Collections WG
> <***@***.***-groups.org> *Date: *10 May 2016, 09:09
>> ... via GoTo meeting?
>>
>>
>> Am 09.05.2016 um 17:51 schrieb TobiasWeigel:
>>> Dear all,
>>>
>>> I've extended Thomas' draft with my part on models. I took up
>>> Frederik's suggestion to look at this as traits as this allowed
>>> me to circumvent category issues and also explain some cases
>>> through combined traits. See attached. Talk to you tomorrow
>>> :-)
>>>
>>> Best, Tobias
>>>
>>> -------- Original Message -------- Subject:
>>> [rda-collection-wg] First draft of specifications From:
>>> ThomasZastrow
<***@***.***> To: RDA Collections
>>> WG <***@***.***-groups.org> Date: 02 May 2016, 15:40
>>>
>>>> Dear all,
>>>>
>>>> So maybe its the best way to send you the first draft of a
>>>> specification document via good old email ;-)
>>>>
>>>> Best,
>>>>
>>>> Tom
>>>>
>>>>
>>>>
>>>> -- Full post:
>>>> https://rd-alliance.org/group/research-data-collections-wg/post/fir
st-draft-specifications
>>>>
>>>>
>>>>
Manage my subscriptions: https://rd-alliance.org/mailinglist
>>>> Stop emails for this post:
>>>> https://rd-alliance.org/mailinglist/unsubscribe/52221
>>>
>>>
>>>
>>> -- Full post:
>>> https://rd-alliance.org/group/research-data-collections-wg/post/firs
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Same problem here...
Am 10.05.2016 um 15:06 schrieb TobiasWeigel:
> .. also wondering: I'm currently using the link for the montly
> call, but not sure if that is going to work as it says "waiting for
> the RDA secretariat"...
>
> -------- Original Message -------- *Subject: *Re:
> [rda-collection-wg] First draft of specifications *From:
> *ThomasZastrow
<***@***.***> *To: *TobiasWeigel
> <***@***.***>, RDA Collections WG
> <***@***.***-groups.org> *Date: *10 May 2016, 09:09
>> ... via GoTo meeting?
>>
>>
>> Am 09.05.2016 um 17:51 schrieb TobiasWeigel:
>>> Dear all,
>>>
>>> I've extended Thomas' draft with my part on models. I took up
>>> Frederik's suggestion to look at this as traits as this allowed
>>> me to circumvent category issues and also explain some cases
>>> through combined traits. See attached. Talk to you tomorrow
>>> :-)
>>>
>>> Best, Tobias
>>>
>>> -------- Original Message -------- Subject:
>>> [rda-collection-wg] First draft of specifications From:
>>> ThomasZastrow
<***@***.***> To: RDA Collections
>>> WG <***@***.***-groups.org> Date: 02 May 2016, 15:40
>>>
>>>> Dear all,
>>>>
>>>> So maybe its the best way to send you the first draft of a
>>>> specification document via good old email ;-)
>>>>
>>>> Best,
>>>>
>>>> Tom
>>>>
>>>>
>>>>
>>>> -- Full post:
>>>> https://rd-alliance.org/group/research-data-collections-wg/post/fir
st-draft-specifications
>>>>
>>>>
>>>>
Manage my subscriptions: https://rd-alliance.org/mailinglist
>>>> Stop emails for this post:
>>>> https://rd-alliance.org/mailinglist/unsubscribe/52221
>>>
>>>
>>>
>>> -- Full post:
>>> https://rd-alliance.org/group/research-data-collections-wg/post/firs
t-draft-specifications
>>>
>>>
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Same problem here...
Am 10.05.2016 um 15:06 schrieb TobiasWeigel:
> .. also wondering: I'm currently using the link for the montly
> call, but not sure if that is going to work as it says "waiting for
> the RDA secretariat"...
>
> -------- Original Message -------- *Subject: *Re:
> [rda-collection-wg] First draft of specifications *From:
> *ThomasZastrow
<***@***.***> *To: *TobiasWeigel
> <***@***.***>, RDA Collections WG
> <***@***.***-groups.org> *Date: *10 May 2016, 09:09
>> ... via GoTo meeting?
>>
>>
>> Am 09.05.2016 um 17:51 schrieb TobiasWeigel:
>>> Dear all,
>>>
>>> I've extended Thomas' draft with my part on models. I took up
>>> Frederik's suggestion to look at this as traits as this allowed
>>> me to circumvent category issues and also explain some cases
>>> through combined traits. See attached. Talk to you tomorrow
>>> :-)
>>>
>>> Best, Tobias
>>>
>>> -------- Original Message -------- Subject:
>>> [rda-collection-wg] First draft of specifications From:
>>> ThomasZastrow
<***@***.***> To: RDA Collections
>>> WG <***@***.***-groups.org> Date: 02 May 2016, 15:40
>>>
>>>> Dear all,
>>>>
>>>> So maybe its the best way to send you the first draft of a
>>>> specification document via good old email ;-)
>>>>
>>>> Best,
>>>>
>>>> Tom
>>>>
>>>>
>>>>
>>>> -- Full post:
>>>> https://rd-alliance.org/group/research-data-collections-wg/post/fir
st-draft-specifications
>>>>
>>>>
>>>>
Manage my subscriptions: https://rd-alliance.org/mailinglist
>>>> Stop emails for this post:
>>>> https://rd-alliance.org/mailinglist/unsubscribe/52221
>>>
>>>
>>>
>>> -- Full post:
>>> https://rd-alliance.org/group/research-data-collections-wg/post/firs
t-draft-specifications
>>>
>>>
Manage my subscriptions: https://rd-alliance.org/mailinglist
>>> Stop emails for this post:
>>> https://rd-alliance.org/mailinglist/unsubscribe/52221
>>
>>
>>
>> -- Full post:
>> https://rd-alliance.org/group/research-data-collections-wg/post/first
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Same problem here...
Am 10.05.2016 um 15:06 schrieb TobiasWeigel:
> .. also wondering: I'm currently using the link for the montly
> call, but not sure if that is going to work as it says "waiting for
> the RDA secretariat"...
>
> -------- Original Message -------- *Subject: *Re:
> [rda-collection-wg] First draft of specifications *From:
> *ThomasZastrow
<***@***.***> *To: *TobiasWeigel
> <***@***.***>, RDA Collections WG
> <***@***.***-groups.org> *Date: *10 May 2016, 09:09
>> ... via GoTo meeting?
>>
>>
>> Am 09.05.2016 um 17:51 schrieb TobiasWeigel:
>>> Dear all,
>>>
>>> I've extended Thomas' draft with my part on models. I took up
>>> Frederik's suggestion to look at this as traits as this allowed
>>> me to circumvent category issues and also explain some cases
>>> through combined traits. See attached. Talk to you tomorrow
>>> :-)
>>>
>>> Best, Tobias
>>>
>>> -------- Original Message -------- Subject:
>>> [rda-collection-wg] First draft of specifications From:
>>> ThomasZastrow
<***@***.***> To: RDA Collections
>>> WG <***@***.***-groups.org> Date: 02 May 2016, 15:40
>>>
>>>> Dear all,
>>>>
>>>> So maybe its the best way to send you the first draft of a
>>>> specification document via good old email ;-)
>>>>
>>>> Best,
>>>>
>>>> Tom
>>>>
>>>>
>>>>
>>>> -- Full post:
>>>> https://rd-alliance.org/group/research-data-collections-wg/post/fir
st-draft-specifications
>>>>
>>>>
>>>>
Manage my subscriptions: https://rd-alliance.org/mailinglist
>>>> Stop emails for this post:
>>>> https://rd-alliance.org/mailinglist/unsubscribe/52221
>>>
>>>
>>>
>>> -- Full post:
>>> https://rd-alliance.org/group/research-data-collections-wg/post/firs
t-draft-specifications
>>>
>>>
Manage my subscriptions: https://rd-alliance.org/mailinglist
>>> Stop emails for this post:
>>> https://rd-alliance.org/mailinglist/unsubscribe/52221
>>
>>
>>
>> -- Full post:
>> https://rd-alliance.org/group/research-data-collections-wg/post/first
- -draft-specifications
>>
>>

Hi group,
thank you, Tobias - this looks like a good start!
To give input to some points made in the discussion below.
1) I agree that persistence is realized on different levels, and identifier, collection and object are the ones I came up with as well.
2) From 1) follows that extendable and shrinkable are mutability on the collection level. “Updatable” would be a term that describes mutability in the objects themselves.
2a) What about reordering an ordered collection?
3) I’d say that we have to keep the source collection in mind as well: pers. rule + pers. source = pers. resulting collection
3a) Generally, I’d think in terms of per reference and per value here. Meaning that the result of the rule is either dynamic or frozen.
X) I might have been unclear here - persistence is certainly not an attribute of collections in general. But in order to manage citability, we’ll have to provide it, I think. My thinking is thus that any changes to something that’s been dubbed persistent create a new version, and each individual version remains persistent to allow for citabillty.
Best,
Frederik

Hi Frederik,
1) Good, so we agree that these should be defined. :-)
2) Yes.
2a) That is a good suggestion for further operations on the 'ordered'
trait. Switch two objects by index, or provide a map old:new.
3) Is a rule-based collection always constructed by giving an existing
collection plus a (set of) rule(s)? I am not sure if this covers all
cases Ulrich may have in mind, though I suspect it may. In this case,
yes, there are some things we can deduce upon construction.
3a) I'd think of the result frozen within the scope/lifetime of a
cursor. An implementation of a rule-based collection, following 3.,
would preferably use a construction by reference (something akin to a
database view). But again, I am not sure if this covers all rule-based
collection ideas completely.
X) Citability is one possible use case, but not the only one. Also: Some
use cases are about citing collections that grow over time, so are at
least 'Extendable'. (Javier's, or in general those who deal with time
series).
Following today's call and what's in the document, I also consider these
limits of scope as agreed:
- We only deal with finite collections.
- We require every collection to have a PID.
- We don't assign PIDs to objects or offer a mechanism for that. We
assume that items already bear PIDs and if they don't, we assume some
form of 'internal' ID. (discussed some weeks earlier)
- The API operations are synchronous.
- The API operations are preferably stateless, but exceptions are
possible regarding query cursors.
- ...?
Might be good to put in the doc somewhere before the meat sections.
Best, Tobias
-------- Original Message --------
*Subject: *Re: [rda-collection-wg] First draft of specifications
*From: *fbaumgardt <***@***.***>
*To: *TobiasWeigel <***@***.***>, Research Data Collections WG
<***@***.***-groups.org>