Missing data in get.first.activity.data leads to an error

[bockthom]

Description

When getting the first activity data via get.first.activity.data, using multiple activity types (activity.type = c("mails", "commits", "issues")),
we end up in an error when there is no activity for several of these activity types.

In particular, assume that there is no "mail" activity in the range data and no "commit" activity in the range data, but only "issue" activity. In such a case, when reducing activity.by.type, we get an error here:

coronet/util-networks-covariates.R

Line 1031 in a656026

result = Reduce(function(x, y) {

In this case, both x ("mails") and y ("commits") are empty lists (list()) in the first step of the reduction, making the reduction fail as there are no names in x and y.

Solutions

Quick and dirty solution:

Check if keys is NULL before

coronet/util-networks-covariates.R

Line 1040 in a656026

result = parallel::mclapply(keys, function(key) {

, and if so, return and empty list instead of combining the values for the keys. Disadvantage: When someone expects that there is a "commits" and "emails" named element in the resulting list for each developer, then this will lead to problems as there is no "commits" or "emails" element in the resulting list.

Preferred solution:

Make sure that the input lists for the reduction (i.e., x and y) are not empty. So, check before the reduction whether multiple lists in activity.by.type are empty, and if there are some, just copy one of the non-empty lists and set all its values to NA. Then reducing the lists should work without any problem and the resulting list in the end should have elements named "commits" and "emails" for each developer though commit data and e-mail data are empty.

Versions

This affects all versions in which the function get.first.activity.data exists.