vo-dml for cube
Gerard Lemson
glemson1 at jhu.edu
Fri Dec 11 17:43:25 CET 2015
Hi Arnold
Am traveling, will reply in more detail later, but a quick over your use cases confirms the point that all can and (in my opinion) SHOULD be modelled NOT as attributes but as collections.
Arguments about code have no place here, VO-DML is supposed to be application neutral.
Serializations can optimize the storage requirements of their representations quite easily.
So I think you need to do better.
Wrt STC progress, I think you need support with that effort indeed, have other people look at it.
I am happy to try to create versions that are semantically at least as rich as what you have now, more reusable, and eminently serializable, whilst staying within the bounds of the current VO-DML doc. But others with more relevant domain knowledge than I MUST start looking at the contents of STC as is.
Cheers
Gerard
From: Arnold Rots [mailto:arots at cfa.harvard.edu]
Sent: Friday, December 11, 2015 10:32 AM
To: Gerard Lemson <glemson1 at jhu.edu>
Cc: CresitelloDittmar, Mark <mdittmar at cfa.harvard.edu>; Omar Laurino (olaurino at head.cfa.harvard.edu) <olaurino at head.cfa.harvard.edu>; msdemlei at ari.uni-heidelberg.de; dm at ivoa.net
Subject: Re: vo-dml for cube
Yes, what I am proposing is an extension of what was agreed regarding the [0..]<literal> specification.
I think requiring that the length of an array of values that is an attribute of an Objecttype (class) or Datatype (structure) has to be fixed by the data model is too restrictive.
There are plenty of use cases where that length can only sensibly be set when instantiating the Object or Data item.
Examples are:
- The array of coefficients in a polynomial class; its length should be set by the order of the polynomial expression, which is a legitimate parameter for such an Objecttype
- The pixel data values in a spectrum
- The coordinate values along an enumerated coordinate axis
- The number of samples in a time series
Just think how often the length of an array is left variable in your code and set at runtime.
VO-DML in its current specification requires that in such situations the array not be modeled as an array of data values (either primitive or compound - like structures), but as lists of objects - the instantiations of classes.
My sense is that that is overkill and results in unnecessary overhead. There is no good reason to have to model a simple coefficient as a class in order to be able to use it in a polynomial class that has its order as a attribute.
I think allowing multiplicity of attributes to be variable, but requiring it to be explicitly specified by a nonnegative integer attribute is an acceptable compromise for this issue.
For me, the matter is that I need to have this issue resolved before I can proceed with the remainder of the STC2 development; it has become a major stumbling block requiring changes that I would prefer not to have to make.
And, at the same time, I believe that this is an issue that will trip up future VO-DML developers as well, so I would rather see it changed right now.
Cheers,
- Arnold
-------------------------------------------------------------------------------------------------------------
Arnold H. Rots Chandra X-ray Science Center
Smithsonian Astrophysical Observatory tel: +1 617 496 7701
60 Garden Street, MS 67 fax: +1 617 495 7356
Cambridge, MA 02138 arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>
USA http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------------------------------------------
On Fri, Dec 11, 2015 at 7:58 AM, Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>> wrote:
HI (adding dm at ivoa in the Cc)
Allowing 0..n, implying EITHER 0 OR exactly n with n a literal is same as discussed before (and below in email trail).
The current doc does not include this yet, I have had no other opinions about this but yours, but I think it is reasonable.
I thought I had already updated the doc, but apparently not yet.
Allowing n to be an integer attribute with multiplicity 1 I am still strongly against.
For example, what if, after initializing the 0..n attribute, you change the value of n?
This possibility implies to me still that the concept represented by the 0..n attribute should not be a datatype, but should be represented by a collection of object types.
I am also still doubtful of the use cases that would seem to require this.
The cases where I have seen you use this feature in STC can and imho SHOULD be designed differently.
I think the dm group MUST start looking seriously at STC before its presumed requirements are used to influence the language.
Any comments from the others.
Cheers
Gerard
From: Arnold Rots [mailto:arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>]
Sent: Thursday, December 10, 2015 5:23 PM
To: Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>>
Cc: CresitelloDittmar, Mark <mdittmar at cfa.harvard.edu<mailto:mdittmar at cfa.harvard.edu>>; Omar Laurino (olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>) <olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>>; msdemlei at ari.uni-heidelberg.de<mailto:msdemlei at ari.uni-heidelberg.de>
Subject: Re: vo-dml for cube
Would it be acceptable extend this slightly:
Multiplicity must be "[0..]n", meaning that the multiplicity is either 0 (but only if "0.." is present) or n,
and where n is either a numeric literal or an existing nonnegative integer attribute with multiplicity 1.
Cheers,
- Arnold
-------------------------------------------------------------------------------------------------------------
Arnold H. Rots Chandra X-ray Science Center
Smithsonian Astrophysical Observatory tel: +1 617 496 7701<tel:%2B1%20617%20496%207701>
60 Garden Street, MS 67 fax: +1 617 495 7356<tel:%2B1%20617%20495%207356>
Cambridge, MA 02138 arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>
USA http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------------------------------------------
On Sat, Oct 31, 2015 at 9:16 AM, Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>> wrote:
Hi Arnold
Thanks, will pass through pipeline.
To all:
I think we should allow 0..n for attributes, but with a different interpretation from its counterpart in composition relations and references.
It would imply that the value of the attribute is either *, or an array of length exactly n.
In relations the interpretation remains that one can have between 0 and n (inclusive) instances in the collection.
Makes sense?
Shall I update docs?
Cheers
Gerard
From: Arnold Rots [mailto:arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>]
Sent: Saturday, October 31, 2015 8:50 AM
To: Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>>
Cc: Omar Laurino (olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>) <olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>>; Mark Cresitello-Dittmar <mdittmar at cfa.harvard.edu<mailto:mdittmar at cfa.harvard.edu>>; msdemlei at ari.uni-heidelberg.de<mailto:msdemlei at ari.uni-heidelberg.de>
Subject: Re: vo-dml for cube
I have checked in my latest version, 2015-10-30.
- Arnold
-------------------------------------------------------------------------------------------------------------
Arnold H. Rots Chandra X-ray Science Center
Smithsonian Astrophysical Observatory tel: +1 617 496 7701<tel:%2B1%20617%20496%207701>
60 Garden Street, MS 67 fax: +1 617 495 7356<tel:%2B1%20617%20495%207356>
Cambridge, MA 02138 arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>
USA http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------------------------------------------
On Thu, Oct 29, 2015 at 8:46 AM, Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>> wrote:
Hi Arnold
(Cc-ing Markus)
It was an unforeseen advantage that VO-DML makes it so much easier identifying errors.
Could you check in your latest version of the Umodel file(s) as well?
In particular because I realize you are still using <<skosconcept>>, whereas my XSLT is now looking for <<semanticconcept>>.
Easy to fix by renaming stereotype. Also rename broadestSKOSConcept to topConcept. I realize I can do the transformation in XSLT, but better if the Umodel profile is as close to VO-DML as possible.
The aspect ratio of the PNG is bit hard to control in GraphViz.
It should be seen on a screen where you can zoom in/out and move the diagram. Hard to print especially for unwieldy models ;)
I find it nice to zoom out a lot to see where what colors reside and which areas have lots of associations.
Is there a document already describing the model? Are you going to add description elements in the UML so they can be picked up by the XSLT?
In particular, I’d be curious for a description of the contents of the frametransform package. And about how/where is it supposed to be used.
Cheers
Gerard
From: Arnold Rots [mailto:arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>]
Sent: Wednesday, October 28, 2015 6:37 PM
To: Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>>
Cc: Omar Laurino (olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>) <olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>>; Mark Cresitello-Dittmar <mdittmar at cfa.harvard.edu<mailto:mdittmar at cfa.harvard.edu>>
Subject: Re: vo-dml for cube
Thank you.
I had missed the error in coords - it's pretty hard to find in UModel.
Yes, four double-l typos
I replaced the subset on PixelFrame by a constraint.
I had made an attempt to see whether providing defaults for attributes like npix
would solve the illegal multiplicities, but that doesn't do the trick. I guess that
one is for you to figure out.
The question about the Qualifier was prompted by my wondering whether
that could/should be the mechanism to split on the basis of dimensionality
(instead of the naxes=naxes constraints).
In specialized I wasn't clear on where to use "=" and where "->".
I know you like those PNGs - I don't, because their ridiculous aspect
ratio makes them very hard to navigate and impossible to print.
Cheers,
- Arnold
-------------------------------------------------------------------------------------------------------------
Arnold H. Rots Chandra X-ray Science Center
Smithsonian Astrophysical Observatory tel: +1 617 496 7701<tel:%2B1%20617%20496%207701>
60 Garden Street, MS 67 fax: +1 617 495 7356<tel:%2B1%20617%20495%207356>
Cambridge, MA 02138 arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>
USA http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------------------------------------------
On Wed, Oct 28, 2015 at 8:42 AM, Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>> wrote:
Hi Arold
Pretty similar, you have now and then some typos (pixellType with double l). Attached the error report.
I had made the coords.SpatialCoord.error an attribute, you still have it as reference for some reason.
You used to have a subsetted role on PixelFrame but no longer.
Thanks for your pointer to qualifier documentation. From description there this seems to be mainly aimed at implementations. If one would like to partition collections of associated objects for conceptual (rather than speed) reasons, you can (and I think “should”) always do that explicitly.
Cheers
Gerard
From: Arnold Rots [mailto:arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>]
Sent: Wednesday, October 28, 2015 7:45 AM
To: Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>>
Cc: Omar Laurino (olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>) <olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>>; Mark Cresitello-Dittmar <mdittmar at cfa.harvard.edu<mailto:mdittmar at cfa.harvard.edu>>
Subject: Re: vo-dml for cube
Gerard,
Yes I got it and, for the sake of understanding it, produced my own version.
Can you check the attached XMI?
I am not sure I have the constraint syntax right in the Specialized package.
The qualifier question isn't really relevant anymore, but no, I didn't mean
ordered. I meant the use of a qualifier on an association, as in
http://www.uml-diagrams.org/property.html
Cheers,
- Arnold
-------------------------------------------------------------------------------------------------------------
Arnold H. Rots Chandra X-ray Science Center
Smithsonian Astrophysical Observatory tel: +1 617 496 7701<tel:%2B1%20617%20496%207701>
60 Garden Street, MS 67 fax: +1 617 495 7356<tel:%2B1%20617%20495%207356>
Cambridge, MA 02138 arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>
USA http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------------------------------------------
On Wed, Oct 28, 2015 at 6:54 AM, Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>> wrote:
Hi Arnold
Yes, when looking at the model I understood your question better.
Have you seen my version with the possible fixes in place?
Btw, when you asked about qualifiers relations, did you think about things like “ordered”?
That particular one already exists (isOrdered), were there other ones you were thinking about?
Btw, In that case the ordering is supposed to be “implicitly” imposed by the order in which elements were added to the collection.
Cheers
Gerard
From: Arnold Rots [mailto:arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>]
Sent: Tuesday, October 27, 2015 4:27 PM
To: Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>>
Cc: Omar Laurino <olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>>; Mark Cresitello-Dittmar <mdittmar at cfa.harvard.edu<mailto:mdittmar at cfa.harvard.edu>>
Subject: Re: vo-dml for cube
Ah, yes. That is what my earlier question was about.
Will try to fix.
- Arnold
-------------------------------------------------------------------------------------------------------------
Arnold H. Rots Chandra X-ray Science Center
Smithsonian Astrophysical Observatory tel: +1 617 496 7701<tel:%2B1%20617%20496%207701>
60 Garden Street, MS 67 fax: +1 617 495 7356<tel:%2B1%20617%20495%207356>
Cambridge, MA 02138 arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>
USA http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------------------------------------------
On Tue, Oct 27, 2015 at 9:43 AM, Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>> wrote:
Hi Arnold
Getting errors below after transforming to VO-DML and validating.
Most have to do with your use of an attribute inside a multiplicity constraint. Please create a real constraint about that in some way.
And there is the case where you define a constraint by susetting, in Xform3D for example. Use an ordinary constraint there.
I can make a suggested version and check it in once I’m at workCHeers
Gerard
XSD:
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:445:54: cvc-pattern-valid: Value '{multiplicity=naxes}' is not facet-valid with respect to pattern '[\w_-]+:[\w_\-/\./*]+' for type 'VODMLREF'.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:445:54: cvc-type.3.1.3: The value '{multiplicity=naxes}' of element 'vodml-ref' is not valid.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:772:53: cvc-pattern-valid: Value '{multiplicity=npix}' is not facet-valid with respect to pattern '[\w_-]+:[\w_\-/\./*]+' for type 'VODMLREF'.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:772:53: cvc-type.3.1.3: The value '{multiplicity=npix}' of element 'vodml-ref' is not valid.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:827:53: cvc-pattern-valid: Value '{multiplicity=npix}' is not facet-valid with respect to pattern '[\w_-]+:[\w_\-/\./*]+' for type 'VODMLREF'.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:827:53: cvc-type.3.1.3: The value '{multiplicity=npix}' of element 'vodml-ref' is not valid.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:882:53: cvc-pattern-valid: Value '{multiplicity=npix}' is not facet-valid with respect to pattern '[\w_-]+:[\w_\-/\./*]+' for type 'VODMLREF'.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:882:53: cvc-type.3.1.3: The value '{multiplicity=npix}' of element 'vodml-ref' is not valid.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:937:53: cvc-pattern-valid: Value '{multiplicity=npix}' is not facet-valid with respect to pattern '[\w_-]+:[\w_\-/\./*]+' for type 'VODMLREF'.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:937:53: cvc-type.3.1.3: The value '{multiplicity=npix}' of element 'vodml-ref' is not valid.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:992:53: cvc-pattern-valid: Value '{multiplicity=npix}' is not facet-valid with respect to pattern '[\w_-]+:[\w_\-/\./*]+' for type 'VODMLREF'.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:992:53: cvc-type.3.1.3: The value '{multiplicity=npix}' of element 'vodml-ref' is not valid.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1047:52: cvc-pattern-valid: Value '{muliplicity=npix}' is not facet-valid with respect to pattern '[\w_-]+:[\w_\-/\./*]+' for type 'VODMLREF'.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1047:52: cvc-type.3.1.3: The value '{muliplicity=npix}' of element 'vodml-ref' is not valid.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1102:52: cvc-pattern-valid: Value '(multiplicity=npix' is not facet-valid with respect to pattern '[\w_-]+:[\w_\-/\./*]+' for type 'VODMLREF'.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1102:52: cvc-type.3.1.3: The value '(multiplicity=npix' of element 'vodml-ref' is not valid.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1207:54: cvc-pattern-valid: Value '(multiplicity=order}' is not facet-valid with respect to pattern '[\w_-]+:[\w_\-/\./*]+' for type 'VODMLREF'.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1207:54: cvc-type.3.1.3: The value '(multiplicity=order}' of element 'vodml-ref' is not valid.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1320:55: cvc-pattern-valid: Value '{mulitiplicity=naxes}' is not facet-valid with respect to pattern '[\w_-]+:[\w_\-/\./*]+' for type 'VODMLREF'.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1320:55: cvc-type.3.1.3: The value '{mulitiplicity=naxes}' of element 'vodml-ref' is not valid.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1509:49: cvc-pattern-valid: Value '{Xform.naxes=1}' is not facet-valid with respect to pattern '[\w_-]+:[\w_\-/\./*]+' for type 'VODMLREF'.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1509:49: cvc-type.3.1.3: The value '{Xform.naxes=1}' of element 'vodml-ref' is not valid.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1550:49: cvc-pattern-valid: Value '{Xform.naxes=2}' is not facet-valid with respect to pattern '[\w_-]+:[\w_\-/\./*]+' for type 'VODMLREF'.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1550:49: cvc-type.3.1.3: The value '{Xform.naxes=2}' of element 'vodml-ref' is not valid.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1616:49: cvc-pattern-valid: Value '{Xform.naxes=3}' is not facet-valid with respect to pattern '[\w_-]+:[\w_\-/\./*]+' for type 'VODMLREF'.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml:1616:49: cvc-type.3.1.3: The value '{Xform.naxes=3}' of element 'vodml-ref' is not valid.
[schemavalidate] C:\workspaces\eclipse-luna-4.4.1\vo-urp-etc\vo-dml-g-vo\models\STC2\2015-10-25\STC2.vo-dml.xml is not a valid XML document
Schematron:
-------
error:
-------
datatype stc2:stctypes.MultiStruct of reference coords.SpatialCoord.error is not an object type but a 'dataType'
From: Arnold Rots [mailto:arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>]
Sent: Tuesday, October 27, 2015 8:24 AM
To: Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>>
Cc: Omar Laurino <olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>>; Mark Cresitello-Dittmar <mdittmar at cfa.harvard.edu<mailto:mdittmar at cfa.harvard.edu>>
Subject: Re: vo-dml for cube
Gerard,
I tried to fix everything up the way it should be
and posted it under 2015-10-25.
Could you see whether that will rattle through your script?
I'm not sure I did get it all right.
I did split the transformations into atomic ones.
- Arnold
-------------------------------------------------------------------------------------------------------------
Arnold H. Rots Chandra X-ray Science Center
Smithsonian Astrophysical Observatory tel: +1 617 496 7701<tel:%2B1%20617%20496%207701>
60 Garden Street, MS 67 fax: +1 617 495 7356<tel:%2B1%20617%20495%207356>
Cambridge, MA 02138 arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>
USA http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------------------------------------------
On Mon, Oct 26, 2015 at 11:52 PM, Arnold Rots <arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>> wrote:
Gerard,
Do I understand correctly that your subsetting constraint is owned by the property to which it applies; while a vanilla constraint is owned by the object that contains the property to which it applies?
Btw, does VO_DML allow qualifiers on relations? I don't think so, but should it?
- Arnold
On Oct 26, 2015 10:35 PM, "Gerard Lemson" <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>> wrote:
Hi Arnold
In UML (MD) I do subsetting still the same way, defining it on the Role (Attribute, Reference or Collection).
(Mark, can you actually do this in Modelio?).
I just now translate it to a SubsettingConstraint on the Type owning the Role rather than defining a separate Role for it. That is a special Constraint, that has a pointer to the subsetted Role.
Has some nice features compared to the old model. Most important is that it solves the problem that Omar identified, that the redefinition would require a separate vo-dml id for the new Role. Also, the concept is really a constraint on allowable values of the structural element defined above.
Cheers
Gerard
From: Arnold Rots [mailto:arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>]
Sent: Monday, October 26, 2015 7:21 AM
To: Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>>
Cc: CresitelloDittmar, Mark <mdittmar at cfa.harvard.edu<mailto:mdittmar at cfa.harvard.edu>>; Omar Laurino (olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>) <olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>>
Subject: Re: vo-dml for cube
Gerard,
How do you want constraints expressed:
just like any ordinary UML constraint, on the class,
or using your subsetting mechanism, in which case it can be associated directly with an attribute?
- Arnold
-------------------------------------------------------------------------------------------------------------
Arnold H. Rots Chandra X-ray Science Center
Smithsonian Astrophysical Observatory tel: +1 617 496 7701<tel:%2B1%20617%20496%207701>
60 Garden Street, MS 67 fax: +1 617 495 7356<tel:%2B1%20617%20495%207356>
Cambridge, MA 02138 arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>
USA http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------------------------------------------
On Sat, Oct 24, 2015 at 9:04 AM, Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>> wrote:
Hi Mark
Ok, I did not import ds as I did not know where the model was.
This can be updated if you point me to it. Do yu have a separate Modelio project for that?
Based on what I did for stc2 you can see how to proceed (I hope ;)
Stereotypes can be created on the root, with “Create stereotype”.
I had problems seeing where they end up.
In case you don’t know this already, on the project view window, in the top right there is a triangle.
Open up the menu and click Show Options > Show MDA model
This will show a new item, “LocalModule” that shows all the stereotypes under LocalProfile.
Right click a stereotype and use Create element > Tag type.
When now assigning the stereotype to an element, you can give values for that tag type.
Select the element, go to the “Elements” Tab. You should see the stereotype show up below UML and Modeler Module.
Clicking it will show the tags and you can add a value.
I am using this for example for modelimport to define the url and documentationURL.
Similarly for <<Model>> you can add author, version etc.
Let me know if yu can see all of this on the Cube-GL.zip version I checked in.
Are you in Australia already? Enjoy the meeting in any case and let me know if/when I may participate in a session.
Cheers
Gerard
From: CresitelloDittmar, Mark [mailto:mdittmar at cfa.harvard.edu<mailto:mdittmar at cfa.harvard.edu>]
Sent: Saturday, October 24, 2015 2:38 AM
To: Gerard Lemson <glemson1 at jhu.edu<mailto:glemson1 at jhu.edu>>
Cc: Arnold Rots <arots at cfa.harvard.edu<mailto:arots at cfa.harvard.edu>>; Omar Laurino (olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>) <olaurino at head.cfa.harvard.edu<mailto:olaurino at head.cfa.harvard.edu>>
Subject: Re: vo-dml for cube
that's great news.
I sent comments from the other thread, but yes.. stc and ds are imported to cube. dataset has it's own model/xmi file.
I'll want to see what you did with the stereotypes.. you were able to generate new ones?
The stc2 model in my cube doc is a little different than arnolds, but most closely related to the 2015-10-14 stuff.
In the end, all the stc2 content will be removed from dataset, and imported as you show.
Mark
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ivoa.net/pipermail/dm/attachments/20151211/3735dbf0/attachment-0001.html>
More information about the dm
mailing list