Bug 334761

Comment 1 Kilian Matt 2011-01-19 08:43:57 EST

The method would look similiar to this:
ChangeSetDiff ScmCore.calculateChanges(ChangeSet base, ChangeSet target, IProgressMonitor monitor);

class ChangeSetDiff {
 ChangeSet base;
 ChangeSet target;

 List<ArtififactDiff> changes; 
}

class ArtifactDiff {
// has a method that can be used to fill a compare editor
 // maybe ChangeType...
}

Comment 2 Steffen Pingel 2011-01-19 23:10:45 EST

My first thought was that a ChangeSet should object represent exactly that. Maybe a ChangeSet should hold onto Changes that describe what happened linking back to Artifacts?

Comment 3 Steffen Pingel 2011-01-19 23:32:15 EST

To be more concrete, I was thinking of something along these lines:

// groups a list of changes identified by the same commit message, author and time
class ChangeSet {
	ScmRepository repository;
	List<Change> changes;
	String message;
	Date date;
	...
}

// describes a change to a single file
class Change {
	ArtifactRevision base;
	ArtifactRevision target;
	ChangeType type;
	...
}

// uniquely identifies a file revision in the repository, similar to IFileRevision
class ArtifactRevision {
	String path;
	String revision;
}

Comment 4 Steffen Pingel 2011-01-19 23:32:43 EST

Created attachment 187165 [details]
model used by the Mylyn Builds project

Comment 5 Kilian Matt 2011-01-20 04:34:29 EST

Yeah that would work and would be something we could use to solve bug 334759 by just providing this model.
In this bug I was proposing a  method for comparing two arbitrary changesets like in svn revision 123 with revision 456, so in that case I would return a list of Change objects - as that method call of comparing two changesets isn't exactly a changeset itself.

Comment 6 Steffen Pingel 2011-01-20 14:46:51 EST

(In reply to comment #5)
> In this bug I was proposing a  method for comparing two arbitrary changesets
> like in svn revision 123 with revision 456, so in that case I would return a
> list of Change objects - as that method call of comparing two changesets isn't
> exactly a changeset itself.

If you are comparing two svn revisions couldn't this be represented with a ChangeSet object? Each change would reference how an artifact changed between the base and target revision. 

If you wanted to get all changes I would expect a method that returns a list of ChangeSets or Change objects given two revisions. Maybe I am not understanding the requirements but unless there is a compelling reason for introducing another model class to represent the differences between two change sets I would lean towards reusing or generalizing the existing classes.

Comment 7 Kilian Matt 2011-01-21 04:51:54 EST

Well I was thinking that a changeset represents exactly one commit in the repository (except that proposed cvs workaround). Using it for a comparision of two different revisions is in my opinion not a good idea. 
It would be better to extract some common interface or base class for a changeset and that compare revision object.

Besides that I could think of one use case for returning a list of all changesets between two revision - the generic implementation for task-changeset mapping would need pretty much that.

Comment 8 Eclipse Webmaster 2022-11-15 11:45:08 EST

Mylyn has been restructured, and our issue tracking has moved to GitHub [1].

We are closing ~14K Bugzilla issues to give the new team a fresh start. If you feel that this issue is still relevant, please create a new one on GitHub.

[1] https://github.com/orgs/eclipse-mylyn