What is the relevence of referenceVersion in Hismo?
Locked
 
|
In the description of Girba, 2005, "Modeling History to understand Software Evolution", there is a small statement on "referenceVersion" in the Hismo metamodel:
"Each Version has a reference to the so called referenceVersion which provides a fixed point in the historical space." This is implemented in Moose Hismo, but there is little description of its relevence to modelling. What is meant by "fixed point in the historical space"? In fact, there is an interface to set it to easily set it to nil!! (HismoEntityVersion class >> with:inHistory:) Can Tudor or someone else able to describe why referenceVersion would be important in time based analysis? Could this be, for example, the "baseline" to which other copies are compared? Can we have a different referenceVersions in different HismoHistoryGroups? Can we change the referenceVersion on the fly during analysis? I can see hints in the names of methods (e.g. isBornPriorToOrInReferenceVersion), but I can't quite catch the semantics. Can you give some guidelines on which version should be selected as the referenceVersion, and why? As I've described earlier, I'm trying to analyse different versions of system development schedules. I'm trying to use the Hismo framework to its best advantage, so I'm trying to understand the semantics as well as the code. Alan. |
Re: What is the relevence of referenceVersion in Hismo?
|
hello
in which context are you doing this task? Stef On Jun 11, 2013, at 11:10 AM, badja <[hidden email]> wrote: > In the description of Girba, 2005, "Modeling History to understand Software > Evolution", there is a small statement on "referenceVersion" in the Hismo > metamodel: > > "Each Version has a reference to the so called referenceVersion which > provides a fixed point in the historical space." > > This is implemented in Moose Hismo, but there is little description of its > relevence to modelling. What is meant by "fixed point in the historical > space"? In fact, there is an interface to set it to easily set it to nil!! > (HismoEntityVersion class >> with:inHistory:) > > Can Tudor or someone else able to describe why referenceVersion would be > important in time based analysis? > > Could this be, for example, the "baseline" to which other copies are > compared? Can we have a different referenceVersions in different > HismoHistoryGroups? Can we change the referenceVersion on the fly during > analysis? I can see hints in the names of methods (e.g. > isBornPriorToOrInReferenceVersion), but I can't quite catch the semantics. > > Can you give some guidelines on which version should be selected as the > referenceVersion, and why? > > As I've described earlier, I'm trying to analyse different versions of > system development schedules. I'm trying to use the Hismo framework to its > best advantage, so I'm trying to understand the semantics as well as the > code. > > Alan. > > > > > -- > View this message in context: http://forum.world.st/What-is-the-relevence-of-referenceVersion-in-Hismo-tp4692748.html > Sent from the Moose mailing list archive at Nabble.com. > _______________________________________________ > Moose-dev mailing list > [hidden email] > https://www.iam.unibe.ch/mailman/listinfo/moose-dev _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
|
|
Not sure on your question: "What context?" I'll try to explain but I may not be addressing what you may be wanting to know.
I have a set of schedule files (MS Project files converted to XML) which are different versions of the same schedule, but updated at different times. My project is Kleio, so I've taken the Hismo metamodel for evolution - (History, Version, Snapshot - from Girba, 2005) and created 3 new classes:
KleioScheduleHistory (subclass of HismoAbstractHistory) (These are analogues to, say, HismoClassHistory, HismoClassVersion and FAMIXClass.) I've simplified this somewhat, as there are also Activities (Tasks) and Resources to consider. These are not important to my question. When I load the schedule data into these classes, I first create one KleioScheduleHistory for the set of schedules (the history). I then create a KleioSchedule for each schedule (the snapshots), and finally I create an instance of KleioScheduleVersion (the versions) for the snapshot using: KleioScheduleVersion class >> with: aVersionEntity inHistory: aHistory withReferenceVersion: aReferenceVersion. This creator method is a modified version of HismoEntityVersion class >> with: aVersionEntity inHistory: aHistory withReferenceVersion: aReferenceVersion. Rather than inherit this creator method from HismoEntityVersion, I've re-written it, as the HismoEntityVersion only works for sub-classes with a name starting with Hismo (see source - 'Hismo' is hard wired as the class prefix!! bad karma). So far, so good. My question however is: What is the significance of the attribute aReferenceVersion in the creator method with:inHistory:withReferenceVersion for the HismoEntityVersion (and so KleioScheduleVersion) class? What problem is it trying to solve? i.e. Why does the HismoEntityVersion creator method have this field as an attribute? I can set it to nil, but what functionality in the Hismo framework would I then be missing? Alan. |
Re: What is the relevence of referenceVersion in Hismo?
|
|
Thanks alan.
Is it a master project? PhD? with which university? On Jun 12, 2013, at 7:32 PM, badja <[hidden email]> wrote: Not sure on your question: "What context?" I'll try to explain but I may not be addressing what you may be wanting to know. _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
|
|
PhD with University of Technology, Sydney - www.uts.edu.au On 13 June 2013 04:19, Stéphane Ducasse <[hidden email]> wrote:
_______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
Re: What is the relevence of referenceVersion in Hismo?
|
|
let us know how we can help. Stef
_______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
|
|
In reply to this post by badja
That sounds really interesting. Is any part of it planned to be open
source?
btw, good to see another Aussie on the list (I'm in Collie Western Australia studying remotely at University of Southern Queensland) cheers -ben badja wrote: Not sure on your question: "What context?" I'll try to explain but I may not be addressing what you may be wanting to know. I have a set of schedule files (MS Project files converted to XML) which are different versions of the same schedule, but updated at different times. My project is Kleio, so I've taken the Hismo metamodel for evolution - (History, Version, Snapshot - from Girba, 2005) and created 3 new classes: KleioScheduleHistory (subclass of HismoAbstractHistory) KleioScheduleVersion (subclass of HismoEntityVersion) KleioSchedule (subclass of MooseEntity) (These are analogues to, say, HismoClassHistory, HismoClassVersion and FAMIXClass.) I've simplified this somewhat, as there are also Activities (Tasks) and Resources to consider. These are not important to my question. When I load the schedule data into these classes, I first create one KleioScheduleHistory for the set of schedules (the history). I then create a KleioSchedule for each schedule (the snapshots), and finally I create an instance of KleioScheduleVersion (the versions) for the snapshot using: KleioScheduleVersion class >> with: aVersionEntity inHistory: aHistory withReferenceVersion: aReferenceVersion. This creator method is a modified version of HismoEntityVersion class >> with: aVersionEntity inHistory: aHistory withReferenceVersion: aReferenceVersion. Rather than inherit this creator method from HismoEntityVersion, I've re-written it, as the HismoEntityVersion only works for sub-classes with a name starting with Hismo (see source - 'Hismo' is hard wired as the class prefix!! bad karma). So far, so good. *My question however is:* What is the significance of the attribute aReferenceVersion in the creator method with:inHistory:withReferenceVersion for the HismoEntityVersion (and so KleioScheduleVersion) class? What problem is it trying to solve? i.e. /*Why*/ does the HismoEntityVersion creator method have this field as an attribute? I can set it to nil, but what functionality in the Hismo framework would I then be missing? Alan. -- View this message in context: http://forum.world.st/What-is-the-relevence-of-referenceVersion-in-Hismo-tp4692748p4693052.html Sent from the Moose mailing list archive at Nabble.com. _______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
|
|
Ben, For what it is worth, all my code will be released under MIT licence:http://ss3.gemstone.com/ss/Kleio.html Alan. On 16 June 2013 16:02, Ben Coman <[hidden email]> wrote:
_______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
Re: What is the relevence of referenceVersion in Hismo?
|
|
In reply to this post by badja
Hi,
Hismo aims to offer you the possibility to navigate both in space and in time. To this end, you need to always know where you are. referenceVersion is similar to mooseModel in FAMIX entities. From any FAMIX entity you can quickly navigate to the model object. In the same way, from any version entity you can navigate to the top-most version (usually a version corresponding to a MooseModel instance). Let's take an example. Imagine you have a history of 5 models (instances of MooseModel) holding FAMIX objects. And now, imagine that the version of a class A in your hand, and you would want to know that it exists in the version 2. So, you would get the object corresponding to system version 2 and then use existsInReferenceVersion: to check your version object (see below):
existsInReferenceVersion: aReferenceVersion "answers if the history has a version in aReferenceVersion" ^(self versions select: [:each | each referenceVersion = aReferenceVersion]) isEmpty not referenceVersion is supposed to be immutable (so you should not change it dynamically). Does it answer the question?
Cheers, Doru On Tue, Jun 11, 2013 at 11:10 AM, badja <[hidden email]> wrote: In the description of Girba, 2005, "Modeling History to understand Software "Every thing has its own flow"
_______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
|
|
Doru,
I'm starting to understand. "Let's take an example. Imagine you have a history of 5 models (instances of MooseModel) holding FAMIX objects." Ah ha. I've been looking at a history and all the versions and snapshots being within a single MooseModel. But you are suggesting I should have a MooseModel for my data at each point in time, then model the time relationships between them using the Hismo history classes. Can I confirm that then the HismoAbtractVersion (as subclassed) and HismoAbstractHistory (as subclassed) instances therefore sit outside the MooseModel instances. The MooseModel instances (in your example 5 instances) would consist of FAMIX class instances, and would not include any Hismo class instances. The history instances (e.g. HismoClassHistory) would point to version instances (HismoClassVersion) which in turn would point to FAMIX class instances (FAMIXClass) - and only these FAMIX instances will be in the (5) MooseModels. Is this correct? |
Re: What is the relevence of referenceVersion in Hismo?
|
|
Hi, You are correct. You need to build a so called HismoModelHistory that will be formed by versions corresponding to your snapshot models. Take a look at MooseSampleData>>initialize for an example of how to initialize this.
Doru On Sun, Jun 23, 2013 at 6:17 AM, badja <[hidden email]> wrote: Doru, "Every thing has its own flow"
_______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
|
|
Doru, Many thanks. I'll let you know how it goes.On 23 June 2013 15:00, Tudor Girba <[hidden email]> wrote:
_______________________________________________ Moose-dev mailing list [hidden email] https://www.iam.unibe.ch/mailman/listinfo/moose-dev |
| Free forum by Nabble | Edit this page |