Version v3.2 #126
Merged
Version v3.2 #126
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
When constructing ranges, in some situations, we might need to avoid that the last range is not a perfect range any more, i.e., it is smaller than the specified time period. Therefore, add a parameter 'imperfect.range.ratio' which quantifies the allowed imperfectness (e.g., at least 70% of the specified time period are needed to keep an imperfect range, otherwise it will be combined with the previous range.) This fixes #104. In addition to that, add a parameter 'include.end.date' to specify whether to add 1 second to the end (for including the end date as the end of ranges is exclusive) or not. (The default should be to add 1 second to the end, as hitherto. However, we came across situations where we explicitlydon't want that behavior, which is the reason for adding the additional parameter.) Also add some additional tests for the newly introduced behavior. Signed-off-by: Thomas Bock <[email protected]>
Signed-off-by: Thomas Bock <[email protected]>
Signed-off-by: Thomas Bock <[email protected]>
To improve the current implementation of the function 'construct.overlapping.ranges' regarding the parameter 'imperfect.range.ratio', we now use one single compact if-block for correcting the ranges *after* their construction. The test suite is adapted to catch the corner case where we obtain a single range which is already shorter than the chosen time period – while passing also 'imperfect.time.ratio' in a manner that would cause the adaptation of the last range. Signed-off-by: Claus Hunsen <[email protected]> Reviewed-by: Thomas Bock <[email protected]>
Signed-off-by: Thomas Bock <[email protected]>
Enable contraction of imperfect ranges in the end Reviewed-by: Claus Hunsen <[email protected]> Reviewed-by: Thomas Bock <[email protected]>
Now, it is possible to construct networks which include more than one artifact and author relation. To provide this functionality, the allowed number of the 'author.relation' and the 'artifact.relation' is changed to Inf instead of 1. A further edge attribute 'relation' is added to describe the relation type. The functions 'get.artifact.network' and 'get.author.network' are changed to handle more than one relation. Therefore, the functions 'get.bipartite.network' and 'get.multi.network' are adjusted, too. The functions use the 'add.edges.for.bipartite.relations' function which can handle more than one relation now. To merge the networks of all relations, we add the functions 'merge.network.data' and 'merge.networks'. The function 'construct.network.from.list' is split into to functions 'construct.edge.list.from.key.value.list' and 'construct.network.from.edge.list'. Solves the second part of #98. Signed-off-by: Barbara Eckl <[email protected]> Signed-off-by: Claus Hunsen <[email protected]>
Now, the colors of the edges depend on the edge attribute 'relation'. The colors of the edges are not any more saved in an edge attribute, because the edge attribute is never been used. Signed-off-by: Barbara Eckl <[email protected]>
To get rid of the global variables in the plot module, we now use the color palette 'viridis' to colorize vertices and edges. The setting of the 'relation' attribute is changed accordingly. Signed-off-by: Claus Hunsen <[email protected]> Signed-off-by: Barbara Eckl <[email protected]>
Now, the different kindes of artifacts (Mail, Issue, File, Function, Feature, Featureexpression, Author) can be plotted with different colors. For this functionality, a new vertex attribute 'kind' is necessary, which includes the information of the attribute 'artifact.type' or 'Author' (for author vertices). The new vertex attribute is respected as vertex color during plotting. Signed-off-by: Barbara Eckl <[email protected]> Signed-off-by: Claus Hunsen <[email protected]>
We would respect the edge attribute 'relation' while simplifying a network, therefore, the new algorithm for simplification is now the following: First, the networks are split depending on the relation type. Then, they are simplified and, in a last step, the networks are merged to one simplified network which have parallel edges if these have different relation types. Also, this works on networks without the 'relation' attribute! Signed-off-by: Barbara Eckl <[email protected]> Signed-off-by: Claus Hunsen <[email protected]>
Signed-off-by: Barbara Eckl <[email protected]>
Remove vertex attribute 'id' of artifact networks with relation 'mail' and add vertex attribute 'artifact.type' in artifact networks of relations 'issues', 'mail' and 'callgraph'. Then these changes of the vertex attributes in artifact networks are included in the test suite. Signed-off-by: Barbara Eckl <[email protected]>
Signed-off-by: Barbara Eckl <[email protected]>
Signed-off-by: Barbara Eckl <[email protected]>
Signed-off-by: Barbara Eckl <[email protected]>
Signed-off-by: Barbara Eckl <[email protected]>
The colors and line types of edges were sometimes wrong. To delete loops from the network before plotting, solves the problem. Signed-off-by: Claus Hunsen <[email protected]> Signed-off-by: Barbara Eckl <[email protected]>
Signed-off-by: Barbara Eckl <[email protected]>
Adding some more comments. Include a new section in 'README.md' for attributes. Signed-off-by: Barbara Eckl <[email protected]>
Signed-off-by: Barbara Eckl <[email protected]>
Signed-off-by: Claus Hunsen <[email protected]>
In this patch, we improve the recently introduced vertex and edge attributes. Details are listed below. Changes regarding vertex attributes: - The attribute 'type' is either TYPE.AUTHOR or TYPE.ARTIFACT. - The attribute 'kind' now contains solely artifact types (such as "Feature" or "Function") and "Author" as values. - The attribute 'artifact.type' is removed completely as it has been partly redundant to the 'kind' attribute. Changes regarding edge attributes: - The attribute 'type' is either TYPE.EDGES.INTRA or TYPE.EDGES.INTER. - The attribute 'artifact.type' is sourced solely from the column 'artifact.type', present in all data sources. - The attribute 'relation' is filled by the NetworkConf values for 'author.relation' or 'artifact.relation', respectively. Additionally, some TODO items are added or adapted. Signed-off-by: Claus Hunsen <[email protected]> Signed-off-by: Barbara Eckl <[email protected]>
To reliably resolve the correct vertex attribute 'kind' for the current artifact relation, we now use a mapping method. This way, we are open to rename any 'kind' value in later steps. Signed-off-by: Claus Hunsen <[email protected]>
As the vertices that we produce in the artifact relation 'mail' represent mail threads, the corresponding vertex kind is renamed to 'MailThread'. Signed-off-by: Claus Hunsen <[email protected]>
As the edges that we produce for the relation 'issue' are initiated by an issue comment – and not the complete issue–, the corresponding artifact type is renamed to 'IssueComment'. Signed-off-by: Claus Hunsen <[email protected]>
For consistent and comprehensible legends for network plots, the legend keys for shapes/linetypes and colors are properly distinguished now. Additionally, the size of the vertex colors are reduced in size to properly match the one of the vertex shapes. Signed-off-by: Claus Hunsen <[email protected]>
Add the parameters 'remove.multiple' and 'remove.loops' to the functions 'simplify.network' and 'simplify.networks' and pass the them to the 'igraph::simplify' function. Signed-off-by: Thomas Bock <[email protected]>
As the edges that we produce for the relation 'issue' are not only initiated by an issue comment, but also by other events, for example opening or closing the issue, the corresponding artifact type is renamed to 'IssueEvent'. Signed-off-by: Barbara Eckl <[email protected]>
Signed-off-by: Barbara Eckl <[email protected]>
Signed-off-by: Barbara Eckl <[email protected]>
As discussed regarding PR #115 and commit 4dead55, the resolution of the vertex kind for artifact networks is now solved through using the available relation variable. This enables us to remove the graph attributes we added earlier. Thanks to @bockthom for pointing this out. Signed-off-by: Claus Hunsen <[email protected]>
To enhance intercompatibility with internal functionality from the package 'igraph', especially with the function 'igraph::as.data.frame(..., what = "both")', the function 'construct.edge.list.from.key.value.list' now returns a data.frame for the vertices, not a plain vector. Signed-off-by: Claus Hunsen <[email protected]>
Add multi-edges to the author and artifact networks Reviewed-by: Thomas Bock <[email protected]> Reviewed-by: Claus Hunsen <[email protected]>
Add revision set id to pasta items Signed-off-by: Christian Hechtl <[email protected]>
Signed-off-by: Christian Hechtl <[email protected]>
Fix bug that throws an error when the core classification is empty Signed-off-by: Christian Hechtl <[email protected]>
Change logging to improve performance Signed-off-by: Christian Hechtl <[email protected]>
Signed-off-by: Christian Hechtl <[email protected]>
Signed-off-by: Christian Hechtl <[email protected]>
Signed-off-by: Christian Hechtl <[email protected]>
Therefor add a method that returns the names if the cached data objects in order to not having to compare all data sources but only the cached ones. This fixes #116. Signed-off-by: Christian Hechtl <[email protected]>
Signed-off-by: Christian Hechtl <[email protected]>
As with the changes in PR #115 the 'viridis' package was added as needed package, also update the install.R. In addition, fix the misspelling of a variable name in install.R. Signed-off-by: Thomas Bock <[email protected]>
Signed-off-by: Christian Hechtl <[email protected]>
This patch contains the following changes: Return `FALSE` when no second `ProjectData` object is given to the equals method Add parantheses to the function calls in the equals method of `ProjectData` Change the revision set id in the pasta reading method from a global parameter to a `seq_along` list and iterate with `mcmapply` Signed-off-by: Christian Hechtl <[email protected]>
Change the revision.set.id from an integer to a string of the form <revision-set-ID> Signed-off-by: Christian Hechtl <[email protected]>
This is because the timeststamps are implicitly tested anyway and they cause problems as the getter of them reads all not cached data sources and this changes the data object that is tested. Signed-off-by: Christian Hechtl <[email protected]>
Signed-off-by: Christian Hechtl <[email protected]>
To do so, fix the check for the object type in the equals method of the ProjectData class to also allow RangeData objects. Call this method from the equals method in RangeData and then compare the remaining two data sources (`range` and `revision.callgraph`). Also add a test for this and fix the faulty documentation of `get.data.cut.to.same.date()` This fixes #116 Signed-off-by: Christian Hechtl <[email protected]>
Add check that one can not compare a ProjectData object to a RangeData object Expand test to that effect Signed-off-by: Christian Hechtl <[email protected]>
Signed-off-by: Christian Hechtl <[email protected]>
Core-Periphery additions and ProjectData object comparison Reviewed-by: Thomas Bock <[email protected]> Reviewed-by: Claus Hunsen <[email protected]>
Signed-off-by: Claus Hunsen <[email protected]>
|
As we reviewed everything before, I will merge rightaway. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
3.2
Changes in detail
v3.1.2...510863e
Added
author.relationandartifact.relationin theNetworkConf(2f1b4d9)relationrepresentingauthor.relationorartifact.relation, respectively, for all types of networks (2f1b4d9)get.bipartite.relation(2f1b4d9)relationduring network simplification (021ac8b)kinddescribing the actual vertex kind (7c628fb, 784c417)merge.networks(2f1b4d9)merge.network.data(2f1b4d9)create.empty.edge.list(2f1b4d9)ProjectData$equals(Identity of ProjectData objects within NetworkBuilder objects #116, 00df306)Changed/Improved
artifact.typedue to inconsistent use ()idfrom artifact vertices due to inconsistent use (7ad49c4)artifact.typefor uniformity ()width = 0.3 + 0.5 * log(weight)(d791df8)construct.network.from.listinto the two functionsconstruct.edge.list.from.key.value.listandconstruct.network.from.edge.list(2f1b4d9)add.edges.for.bipartite.relation(2f1b4d9)relationduring network simplification (021ac8b)message.idbeing mapped to acommit.hash(992ddf8)revision.set.idto PaStA data to indicate which e-mails are concerned with the same patch (992ddf8)Fixed
ProjectConfsetter in theProjectDataclass really is a object of typeProjectConf(ab00c96)