- The email analysis should be able to run even without a VCS analysis
- We may also wish to explore the mail communication that preceeds
a release range or following a release
Signed-off-by: Mitchell Joblin <mitchell.joblin.ext@xxxxxxxxxxx>
---
codeface/R/ml/analysis.r | 57 +++++++++++++++++++++++++-----------------------
1 file changed, 30 insertions(+), 27 deletions(-)
diff --git a/codeface/R/ml/analysis.r b/codeface/R/ml/analysis.r
index 214d4f2..6c37903 100644
--- a/codeface/R/ml/analysis.r
+++ b/codeface/R/ml/analysis.r
@@ -348,38 +348,41 @@ dispatch.all <- function(conf, repo.path, resdir) {
release.intervals <- release.intervals[nonempty.release.intervals]
release.labels <- release.labels[nonempty.release.intervals]
+ ## Obtain a unique numerical ID for the mailing list (and clear
+ ## any existing results on the way)
+ ml.id <- gen.clear.ml.id.con(conf$con, conf$listname, conf$pid)
+
if (length(nonempty.release.intervals) == 0) {
- stop("Mailing list does not cover any release range.")
+ loginfo("Mailing list does not cover any release range.")
}
+ else {
+ ## TODO: Find some measure (likely depending on the number of messages per
+ ## time) to select suitable time intervals of interest. For many projects,
+ ## weekly (and monthly) are much too short, and longer intervals need to
+ ## be considered.
+ periodic.analysis <- FALSE
+ if (periodic.analysis) {
+ loginfo("Periodic analysis", logger="ml.analysis")
+ ## Select weekly and monthly intervals (TODO: With the new flexible
+ ## intervals in place, we could select proper monthly intervals)
+ iter.weekly <- gen.iter.intervals(dates, 1)
+ iter.4weekly <- gen.iter.intervals(dates, 4)
+
+ analyse.sub.sequences(conf, corp.base, iter.weekly, repo.path, resdir,
+ paste("weekly", 1:length(iter.weekly), sep=""))
+ analyse.sub.sequences(conf, corp.base, iter.4weekly, repo.path, resdir,
+ paste("4weekly", 1:length(iter.4weekly), sep=""))
+ }
- ## TODO: Find some measure (likely depending on the number of messages per
- ## time) to select suitable time intervals of interest. For many projects,
- ## weekly (and monthly) are much too short, and longer intervals need to
- ## be considered.
- periodic.analysis <- FALSE
- if (periodic.analysis) {
- loginfo("Periodic analysis", logger="ml.analysis")
- ## Select weekly and monthly intervals (TODO: With the new flexible
- ## intervals in place, we could select proper monthly intervals)
- iter.weekly <- gen.iter.intervals(dates, 1)
- iter.4weekly <- gen.iter.intervals(dates, 4)
-
- analyse.sub.sequences(conf, corp.base, iter.weekly, repo.path, resdir,
- paste("weekly", 1:length(iter.weekly), sep=""))
- analyse.sub.sequences(conf, corp.base, iter.4weekly, repo.path, resdir,
- paste("4weekly", 1:length(iter.4weekly), sep=""))
- }
+ loginfo("Analysing subsequences", logger="ml.analysis")
- loginfo("Analysing subsequences", logger="ml.analysis")
- ## Obtain a unique numerical ID for the mailing list (and clear
- ## any existing results on the way)
- ml.id <- gen.clear.ml.id.con(conf$con, conf$listname, conf$pid)
+ ## Also obtain a clear plot for the mailing list activity
+ activity.plot.name <- str_c(conf$listname, " activity")
+ activity.plot.id <- get.clear.plot.id(conf, activity.plot.name)
+ analyse.sub.sequences(conf, corp.base, release.intervals, repo.path,
resdir,
+ release.labels, ml.id, activity.plot.id)
- ## Also obtain a clear plot for the mailing list activity
- activity.plot.name <- str_c(conf$listname, " activity")
- activity.plot.id <- get.clear.plot.id(conf, activity.plot.name)
- analyse.sub.sequences(conf, corp.base, release.intervals, repo.path, resdir,
- release.labels, ml.id, activity.plot.id)
+ }
## #######
## Global analysis
--
2.1.4
