[OSLC-RM] Link types
Benjamin Williams
bwilliams at uk.ibm.com
Wed Oct 7 10:40:30 EDT 2009
I agree with Torge that the ability to constrain source and target objects
to specific types is something that is inevitably needed in the long run.
That being said, any such typed link, is, at the end of the day, a child
of the general unconstrained link type, which is a necessary pre-requisite
before any typed links are defined or implemented.
I have just seen a mail come in from Simon that suggests loosely typed
links in the form of labels for 1.0.
I like this approach as it would be easy to evolve into a more strongly
typed representation at a later date.
/Ben
-----------------------------------------------------------------------------
Benjamin Williams
Senior Product Manager - Rational Publishing Engine
Email: bwilliams at uk.ibm.com
Tel: +44 20 8818 4360
Cell: +44 7710 637 067
IBM Extension: 364360
IBM ITN: 37364360
-----------------------------------------------------------------------------
From:
Torge Kummerow <torgato at gmx.de>
To:
oslc-rm at open-services.net
Date:
07/10/2009 13:55
Subject:
Re: [OSLC-RM] Link types
Sent by:
oslc-rm-bounces at open-services.net
Well, if it should be part of V1.0 or not is not my decision to make,
however I would say this is an inevitable future for LinkTypes. To be
able to constrain them to a specific source and target element group.
Especially if we introduce a catalog of LinkTypes this would act as a
good filter, showing only matching Types. However I doubt that a
granularity beyond the domain is of much use. But this depends on the
number of distinguishable sub elements in each domain I guess.
Torge.
James Conallen schrieb:
> I too think that we need to accept for the fact that now each group will
> have its own solution, (which are somewhat similar when you consider the
> potential). I also think that no matter what is done, clients will have
> to change in the future, so why not do something now to support the
> essential use cases and get us all started.
>
> In your second point, you describe a potential property of a link type
> definition. The ability for a specific link type to restrict what can be
> on either end based on the 'domain' of the resource, or maybe even the
> specific type of resource. I worry about getting this aspect of link
> definition right so early on. While I think a generic and flexible link
> management system can be defined for 1.0, I think this particular aspect
> of a link type's definition might not be necessary or 1.0.
>
> <jim/>
>
> jim conallen
> jconallen at us.ibm.com
> Rational Software, IBM Software Group
>
>
>
> Inactive hide details for Torge Kummerow ---10/07/2009 08:21:40
> AM---Well, it can?t hurt, if each group is presenting a solutioTorge
> Kummerow ---10/07/2009 08:21:40 AM---Well, it can?t hurt, if each group
> is presenting a solution it thinks is the best approach. The con
>
>
> From:
> Torge Kummerow <torgato at gmx.de>
>
> To:
> oslc-rm at open-services.net
>
> Date:
> 10/07/2009 08:21 AM
>
> Subject:
> Re: [OSLC-RM] Link types
>
> Sent by:
> oslc-rm-bounces at open-services.net
>
> ------------------------------------------------------------------------
>
>
> ic
> Well, it can?t hurt, if each group is presenting a solution it thinks is
> the best approach. The consolidation will have to take place IHMO and
> then as lots of input for finding the best solution.
>
> I also think that not only the linking spec should be defined globally,
> but also the mechanism in doing so shouldn't care which objects are
> being linked right now, only the LinkTypes should care. (An RM->RM
> LinkType can?t link a QM object with an CM object for example)
>
> Greetings,
> Torge
>
> James Conallen schrieb:
> > I absolutely agree. A uniform linking approach is critical to cross
> > domain collaboration.
> >
> > The problem is that most of the new OSLC workgroups are on the hook
to
> > provide a 1.0 specification by the end of this calendar year. The
> > governing philosophy of the OSLC has been to deliver small sets of
> > specifications that provide value immediately, and that can be built
on
> > later.
> >
> > With that said we can either skip any attempt to define how links can
be
> > managed, since it will most likely change once things are
consolidated
> > later (the OSLC is working on providing common specifications across
all
> > domains). Or we can provide something that is useable now, but will
most
> > likely have to be changed later.
> >
> > If we don't do anything now, the spec will be incomplete with respect
to
> > the essential use cases, since it will require additional knowledge
of
> > the implementing system to be able to manage links, and that is
> > essentially another form of point-to-point integration that we see
> > across the tools today. If we do specify either a flexible approach,
or
> > just define a fixed set of what we consider common link types, then
> > clients built to this spec will have to change (possibly losing
> > functionality) when the later ones come out.
> >
> > <jim/>
> >
> > jim conallen
> > jconallen at us.ibm.com
> > Rational Software, IBM Software Group
> >
> >
> >
> > Inactive hide details for "Simon Wills" ---10/07/2009 07:47:09
> > AM---Hi"Simon Wills" ---10/07/2009 07:47:09 AM---Hi
> >
> >
> > From:
> > "Simon Wills" <simon.wills at integrate.biz>
> >
> > To:
> > "Ian Green1" <ian.green at uk.ibm.com>, James
> Conallen/Philadelphia/IBM at IBMUS
> >
> > Cc:
> > <oslc-rm at open-services.net>
> >
> > Date:
> > 10/07/2009 07:47 AM
> >
> > Subject:
> > RE: [OSLC-RM] Link types
> >
> >
------------------------------------------------------------------------
> >
> >
> >
> > Hi
> >
> > I think that there are three distinct issues here, namely:
> >
> > a. Whether there is a need for a uniform approach to
> > defining link-related resources across OSLC;
> > b. Whether we attempt to define a minimal set of
> > standardised link types;
> > c. Assuming that we **do** adopt a uniform approach to
> > linking across OSLC, how and when we do so.
> >
> > Of these points, deciding whether or not to implement a uniform
linking
> > model is by far and away the most important. Standardising link
labels
> > is a 'nice to have' that follows on from this first decision (this
was
> > the main focus of the RM discussion document), but is something that
we
> > can live without. And the last point is to do with implementation.
> >
> > Ben's interests (i.e. reporting) provide a good illustration of the
key
> > issue. If I understand correctly, he wants to be able to generate
> > reports about lifecycle artefacts, some of which may require him to
> > traverse links between artefacts, without having to make allowances
for
> > the domains in which those artefacts live. This seems to me to be
> > entirely reasonable -- and reporting isn't the only domain for which
> this
> > expectation holds.
> >
> > OSLC stands for Open Services for Lifecycle Collaboration. Lifecycle
> > collaboration requires that we leverage relationships between
lifecycle
> > artefacts. If we don't tackle the issue of uniformity in the
handling
> > of links, we risk ending up with a series of poorly connected domain
> > islands, each with its own domain-specific linking paradigm. Surely
> > there is a /prima facie/ case to recognise that links are the things
> > that tie these domains together, and hence that the definition of
> > link-related resources and their behaviour should be common across
all
> > OSLC domains?
> >
> > Best regards
> >
> > Simon
> >
> > *From:* oslc-rm-bounces at open-services.net
> > [mailto:oslc-rm-bounces at open-services.net] *On Behalf Of *Ian Green1*
> > Sent:* 07 October 2009 11:39*
> > To:* James Conallen*
> > Cc:* oslc-rm at open-services.net*
> > Subject:* Re: [OSLC-RM] Link types
> >
> >
> > Hello Jim,
> >
> > I think your description of how OSLC links currently work in the
> > defining of a relationship is accurate. There has been some
discussion
> > about the merit of such a design, since it is exposed to a fragility,
> > namely, that for each link, there should be a corresponding link in
the
> > other direction. The need for the "backlink" as it has been called,
aids
> > navigability (we typically want the relationship to be navigable in
both
> > directions). It is also reasonably easy to implement.
> >
> > The fragility is maintaining this invariant: for example, when a
> > resource is deleted, any "backlinks" to that resource should be
removed.
> > There are other "relationship management" designs which don't have
this
> > weakness.
> >
> > My reading of Simon's document is that it is talking about these
> > relationships, rather than the lower-level "nuts and bolts" of how
one
> > resource links to some other resource. We need to think about how we
can
> > uniformly manage these relationships across all the domains.
> >
> > best wishes,
> > -ian
> >
> > ian.green at uk.ibm.com (Ian Green1/UK/IBM at IBMGB)
> > Chief Software Architect, Requirements Definition and Management
> > IBM Rational
> >
> > From: James Conallen <jconallen at us.ibm.com>
> > To: Benjamin Williams/UK/IBM at IBMGB
> > Cc: Ian Green1/UK/IBM at IBMGB, oslc-rm at open-services.net
> > Date: 06/10/2009 16:44
> > Subject: Re: [OSLC-RM] Link types
> >
> >
------------------------------------------------------------------------
> >
> >
> >
> > I have one concern about the link types document's attempt to define
> > *both* the forward and reverse links. I believe that the general OSLC
> > mechanism for creating bi-directional links is to create two
> > unidirectional links managed by each server, and hope that it is
> > possible with a query on the other server to find the right back link
> > when deleting.
> >
> > If this is the case, then we should probably restrict the definition
of
> > the link types to those that are managed by the OSLC RM implementing
> > service, and not make assumptions that all the other services (i.e.
OSLC
> > CM, OSLC AM, OSLC QM, ...) will support these OSLC RM defined link
types.
> >
> > In the OSLC AM space our scenarios have all sorts of link types that
are
> > referenced. For example a model element can implement a requirement,
or
> > specify a requirement, or a requirement resource can be documentation
> > (i.e. how to guide) for a model element. Given that both models and
> > requirements can be used in many different ways, it is hard to pin
down,
> > generically, just a few relationship types between AM resources and
> > requirements. With this said, if we just say that links from
> > Requirements to Architecture Management Resources will be one of
> > (modeledBy, realisedBy or incorporatedBy) that will be fine from an
OSLC
> > AM client point of view.
> >
> > <jim/>
> >
> > jim conallen
> > jconallen at us.ibm.com
> > Rational Software, IBM Software Group
> >
> >
> >
> > Inactive hide details for Benjamin Williams ---10/06/2009 10:22:09
> > AM---Ian As per my other email on reporting, I feel that sucBenjamin
> > Williams ---10/06/2009 10:22:09 AM---Ian As per my other email on
> > reporting, I feel that such such collaboration
> >
> >
> > From:
> > Benjamin Williams <bwilliams at uk.ibm.com>
> >
> > To:
> > Ian Green1 <ian.green at uk.ibm.com>
> >
> > Cc:
> > oslc-rm at open-services.net, oslc-rm-bounces at open-services.net
> >
> > Date:
> > 10/06/2009 10:22 AM
> >
> > Subject:
> > Re: [OSLC-RM] Link types
> >
> > Sent by:
> > oslc-rm-bounces at open-services.net
> >
> >
------------------------------------------------------------------------
> >
> >
> >
> > Ian
> >
> > As per my other email on reporting, I feel that such such
collaboration
> > and consistency across domains will be extremely important, if not
> > absolutely necessary.
> > Specifically in the context of reporting, the features exposed by
> > service providers across each domain will need to (consistently)
support
> > the requirements and use-cases defined in the Reporting domain such
that
> > reporting consumers can leverage lifecycle data without needing any
> > knowledge of the source domains and any differences in the way they
> > expose data.
> >
> > The scenarios and use-cases that we are defining in the reporting
domain
> > should be entirely domain agnostic, as far as I am concerned. We can
use
> > specific examples for both single domain reporting as well as cross
> > domain reporting, but the domains used in the examples should be
freely
> > interchangeable without any impact on the reporting consumer.
> >
> > The reporting domain is still in its infancy, and my involvement in
OSLC
> > is only just beginning, but from what I understand, the reporting
domain
> > might be one of the first things to force us to think about some of
the
> > cross domain issues that you and Simon have discussed.
> >
> > /Ben
> >
> >
>
-----------------------------------------------------------------------------
> > Benjamin Williams
> > Senior Product Manager - Rational Publishing Engine
> >
> > Email: bwilliams at uk.ibm.com
> > Tel: +44 20 8818 4360
> > Cell: +44 7710 637 067
> > IBM Extension: 364360
> > IBM ITN: 37364360
> >
>
-----------------------------------------------------------------------------
>
> >
> >
> > From: Ian Green1/UK/IBM at IBMGB
> > To: "Simon Wills" <simon.wills at integrate.biz>
> > Cc: oslc-rm at open-services.net
> > Date: 06/10/2009 14:16
> > Subject: Re: [OSLC-RM] Link types
> > Sent by: oslc-rm-bounces at open-services.net
> >
> >
------------------------------------------------------------------------
> >
> >
> >
> >
> > Hello Simon,
> >
> > this is good progress both on explaining our current position w.r.t.
> > link types, and also taking a longer term view of the role we see
link
> > types playing the future. Thanks very much for putting this together.
> >
> > Steve/Scott: whilst there is no pressure as yet to drive this
forward,
> > Simon makes a good case for any such effort to be cross-OSLC domains.
> > How do you see this shaping up over time? Is there some tension
between
> > designing an OSLC-wide vocabulary and being entirely scenario-driven?
> >
> > best wishes,
> > -ian
> >
> > ian.green at uk.ibm.com (Ian Green1/UK/IBM at IBMGB)
> > Chief Software Architect, Requirements Definition and Management
> > IBM Rational
> >
> > From: "Simon Wills" <simon.wills at integrate.biz>
> > To: Ian Green1/UK/IBM at IBMGB
> > Date: 05/10/2009 15:42
> > Subject: Link types
> >
> >
> >
------------------------------------------------------------------------
> >
> >
> >
> > Hi Ian
> >
> > Just to let you know ... I've put a new section titled 'Discussion
> > Topics' in the right hand sidebar of the main RM page, and put a link
to
> > a discussion document on link types (at
> > _http://open-services.net/bin/view/Main/RmDiscussionLinkTypes_).
Haven't
> > quite sussed out how to format tables properly in the Wiki editor,
but
> > the content is all there.
> >
> > Would you like me to put out an announcement on the mailing list?
> >
> > Cheers
> >
> > Simon
> >
> >
------------------------------------------------------------------------
> >
> > Simon Wills
> > Managing Director *
> > integrate systems engineering ltd*
> > m: +44 (0)7967 091824
> > t: +44 (0)1225 859991
> > f: +44 (0)1225 859993
> > e: simon.wills at integrate.biz
> > w: _www.integrate.biz_ <http://www.integrate.biz/>
> >
------------------------------------------------------------------------
> >
> >
> >
> >
> >
------------------------------------------------------------------------
> >
> > /Unless stated otherwise above:
> > IBM United Kingdom Limited - Registered in England and Wales with
number
> > 741598.
> > Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire
> PO6 3AU/
> >
> >
> >
> >
> >
> > _______________________________________________
> > OSLC-RM mailing list
> > OSLC-RM at open-services.net_
> >
__http://open-services.net/mailman/listinfo/oslc-rm_open-services.net_
> >
> >
> >
------------------------------------------------------------------------
> >
> > /Unless stated otherwise above:
> > IBM United Kingdom Limited - Registered in England and Wales with
number
> > 741598.
> > Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire
> PO6 3AU/
> >
> >
> >
> >
> >
> > _______________________________________________
> > OSLC-RM mailing list
> > OSLC-RM at open-services.net_
> >
__http://open-services.net/mailman/listinfo/oslc-rm_open-services.net_
> >
> >
> >
> >
------------------------------------------------------------------------
> >
> > /Unless stated otherwise above:
> > IBM United Kingdom Limited - Registered in England and Wales with
number
> > 741598.
> > Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire
> PO6 3AU/
> >
> >
> >
> >
> >
> >
------------------------------------------------------------------------
> >
> > _______________________________________________
> > OSLC-RM mailing list
> > OSLC-RM at open-services.net
> > http://open-services.net/mailman/listinfo/oslc-rm_open-services.net
>
> _______________________________________________
> OSLC-RM mailing list
> OSLC-RM at open-services.net
> http://open-services.net/mailman/listinfo/oslc-rm_open-services.net
>
>
_______________________________________________
OSLC-RM mailing list
OSLC-RM at open-services.net
http://open-services.net/mailman/listinfo/oslc-rm_open-services.net
Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number
741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://open-services.net/pipermail/oslc-rm_open-services.net/attachments/20091007/fc526ab4/attachment-0003.html>
More information about the Oslc-Rm
mailing list