new taskview_helper using number of commits about projmgr HOT 6 CLOSED

@konradmayer Wow - thank you so much! I love this idea. I had actually worried a bit that the taskboard_helpers weren't necessarily a particularly good fit for real-world workflows. Now that you mention tracking the commits, this seems like such a clear winner for an option! I also really appreciate the code example. The illustration really helps.

On the implementation side, I'm torn between two instincts. I love the clean syntax of has_n_commits( count = 1) since it's very literate and most in line with the other taskboard_helpers. However, I also want users to do as little work as possible, and I hate to require them to join their own data. Admittedly, I don't have a particularly good reason for this.

The only alternative I can think of is using the syntax has_n_commits( count = 1, events ) and relying on the user to provide the events dataframe (or list) as a prerequisite. Similar to your get_n_commit function, perhaps this means adding a get_issue_set_events(ref, issues) function which could either take the output of get_issues, the output of parse_issues or a vector of issue numbers and return all the events for that set of issues (iterating over get_issue_events() as you did.) That sort of function might be of general use, anyway, since the GitHub API unfortunately doesn't allow for any query parameters on Issue Events.

So, in summary, I guess the two potential workflows could be:

my_issues <- get_issues(my_repo, state = "all", since = "2019-01-01") %>% parse_issues()
my_issues_events <- get_issues_set_events(my_repo, my_issues)
viz_taskboard(my_issues, has_n_commits(3, my_issues_events))

or

my_issues <- get_issues(my_repo, state = "all", since = "2019-01-01") %>% parse_issues()
issues <- issues %>% mutate(n_commits = get_n_commits(my_repo, number))
viz_taskboard(my_issues, has_n_commits(3))

Do you have any opinion on which seems more appealing / usable?

Thank you again so much for this great idea!

from projmgr.

FWIW I discovered a quirk with the event data that I'll stash here for future reference. Commit IDs that close issues show up both in a "commit" event and a "closed" event. For exact counts, we may have to do a distinct count or filter non-null commit IDs to "referenced" events only:

library(projmgr)
library(dplyr)

projmgr <- create_repo_ref('emilyriederer', 'projmgr')
#> Requests will authenticate with GITHUB_PAT
issue_events <- get_issue_events(projmgr, 9)

sum(  vapply( issue_events, function(x) !is.null( x[["commit_id"]] ) , logical(1) ) )
#> [1] 3
sum(  vapply( issue_events, function(x) x[["event"]] == "referenced" , logical(1) ) )
#> [1] 2
sum(  vapply( issue_events, function(x) x[["event"]] == "referenced" & !is.null(x[["commit_id"]]) , logical(1) ) )
#> [1] 2

from projmgr.

(As a sidenote, since it sounds like you might have been looking at viz_taskboard, may I please point out the report_taskboard alternative? I think it's new since you installed the package.

I'm experimenting with HTML-based visualization since the linking workflow is more seamless than ggplot and it also dynamically scales better to page size. The downside, of course, is that it only works in HTML output.

No pressure to look, but I certainly welcome any feedback on whether or not this seems useful compared to the plots!)

from projmgr.

Hey @konradmayer - I just got around to implementing this with a has_n_commits(events, n = 1) function. Per my thoughts above, it does require the user to construct the events dataframe first. Closing this for now, but if you're interested to check it out, I'd love to know what you think!

from projmgr.

Hi @emilyriederer, thanks for further polishing your package!
I tried your new taskview helper, but somehow don't manage to get my head around it.

When I use the function like that,

my_issues <- get_issues(my_repo, state = "all", since = "2019-01-01") %>% parse_issues()
my_events <- get_issue_events(my_repo, 14) %>% parse_issue_events()
viz_taskboard(my_issues, has_n_commits(my_events, n = 1))

I get the same result only showing issue 1 in the "in progress" category, no matter what issue number is chosen in get_issue_events(), if at minimum a single commit is contained in the issue (an error as "no rows to aggregate" otherwise). This output is accompanied by a warning:

Warning message: In if (res != "") { : the condition has length > 1 and only the first element will be used
as res is a longer list.

It seems I don't really understand the intended workflow, as for my_events I would assume I need a dataframe containing the referenced events of all the issues and not of a selected one, so something like the not working code below:

my_issues <- get_issues(my_repo, state = "all", since = "2019-01-01") %>% parse_issues()

my_events <- 
    map(my_issues$number, ~get_issue_events(my_repo, .x) %>% 
        parse_issue_events()) %>% 
        bind_rows()

viz_taskboard(my_issues, has_n_commits(my_events, n = 1))

Therefore I would expect a function similar to get_issue_events() without a number argument to construct the events dataframe. Could you point me to what I'm failing to understand?

from projmgr.

Hi @konradmayer -- Apologies for the slow reply and thank you for your kind and thoughtful reply. I had a bit of a conceptual mistake which made its way both into the has_n_commits() code and the unit tests. So, everything superficially looked fine but the function was just simply wrong.

Even worse than that, you are completely right about the core behavior of get_issue_events() not facilitating this workflow. I have also changed the default behavior there to return a dummy "existence" event for issues with no real events.

All should be working appropriately now. The reprex below is based on the current state of my emilyriederer/experigit repo which currently has two open issues: one with one commit and one with no commits.

library(projmgr)

repo <- create_repo_ref("emilyriederer", "experigit")
#> Requests will authenticate with GITHUB_PAT
issues <- get_issues(repo, state = "open") %>% parse_issues()
events <- purrr::map_df(issues$number, ~get_issue_events(repo, .x) %>% parse_issue_events())
has_n_commits(events, n = 0)(issues)
#> [1] TRUE TRUE
has_n_commits(events, n = 1)(issues)
#> [1]  TRUE FALSE
has_n_commits(events, n = 2)(issues)
#> [1] FALSE FALSE
viz_taskboard(issues, in_progress_when = has_n_commits(events))

^{Created on 2019-09-01 by the reprex package (v0.3.0)}

Thank you again so much for your patience and helpful feedback! It really means a lot.

from projmgr.