Remove empty artifact as vertex (PR #149)

[clhunsen]

Description

With the proposed changes in PR #149 (see this comment), there is an issue when constructing artifact networks while not filtering out empty/untracked artifacts (see MWE below): The empty/untracked artifact is added as a vertex to the artifact networks!

Steps to Reproduce or Minimal Working Example (MWE)

Before running the code below, I checked out the code of PR #149!

git fetch upstream pull/149/head:pr149
git checkout pr149

Then, I added the following line to the file tests/codeface-data/results/testing/test_feature/feature/commits.list:

42;"2018-01-01 15:00:00";"Claus";"[email protected]";"2018-01-01 15:00:00";"Claus";"[email protected]";"foobar5c523d835459c42f33e863623138555e25";1;1;0;1;"";"";"";0

Afterwards, I ran the following code:

## / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /
## Initialization ----------------------------------------------------------

source("util-init.R")

## / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /
## Case-study configuration ------------------------------------------------

CF.DATA = "tests/codeface-data" # path to codeface data

CF.SELECTION.PROCESS = "testing" # releases, threemonth(, testing)

CASESTUDY = "test"
ARTIFACT = "feature" # function, feature, file, featureexpression (only relevant for cochange)

AUTHOR.RELATION = "mail" # mail, cochange, issue
ARTIFACT.RELATION = "cochange" # cochange, callgraph, mail, issue


## / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /
## Project and network configuration ---------------------------------------

## initialize project configuration
proj.conf = ProjectConf$new(CF.DATA, CF.SELECTION.PROCESS, CASESTUDY, ARTIFACT)
proj.conf$update.value("commits.filter.base.artifact", FALSE)
proj.conf$update.value("commits.filter.untracked.files", FALSE)
# proj.conf$print()

## initialize network configuration
net.conf = NetworkConf$new()
net.conf$update.values(updated.values = list(
    author.relation = AUTHOR.RELATION,
    artifact.relation = ARTIFACT.RELATION
))

## construct data object and network (builder)
x.data = ProjectData$new(project.conf = proj.conf)
x = NetworkBuilder$new(project.data = x.data, network.conf = net.conf)

print(
    UNTRACKED.FILE.EMPTY.ARTIFACT %in% igraph::V(x$get.artifact.network())$name
)

Expected output: [1] FALSE

Actual output: [1] TRUE

Reproduces how often: Every time.

Versions

This is only relevant to the changes from PR #149.

Additional Information

We definitely need a test case for this in tests/test-networks-artifact.R, i.e., adding commit data on untracked files to the commit list of the test suite (as proposed above). In particular, we need a line of commit information for the test cases (in the files tests/codeface-data/results/testing/test_feature/feature/commits.list and
tests/codeface-data/results/testing/test_proximity/proximity/commits.list and the corresponding range-folder data, respectively).

Note: We should aim for filtering untracked files in most test cases (which is the default for now) so that we do not need to change that many tests. 😉

[clhunsen]

A short addendum/clarification: We need to talk about proper tests for the changes in PR #149 (involving all new/changed configuration parameters), then we automatically revisit this issue. 😉

[clhunsen]

As PR #158 is merged now, I will close this issue.