data.table work diary: Jul 29 - Aug 2

• July 29, ‘24:

1) Collected feedback on my slides (starting to make new ones accordingly, and learning more of Google slides too), Zoom meetings. (8 hours)

(Going with a wintry theme as it looks aesthetically pleasing to me :)

• July 30, ‘24:

1) Made a performance test for forder caching:

  # Fixed in: https://github.com/Rdatatable/data.table/pull/4386
  "forder improved in #4386" = atime::atime_test(
    N = 10^seq(1, 8), # 1e9 exceeds the runner's memory (process gets killed)
    setup = {
      dt <- data.table(id = sample(N), value = rnorm(N))
    },
    expr = {
      data.table:::setkey(dt, id)
      dt[order(data.table:::forder(dt, "id"))]
      dt[order(data.table:::forder(dt, "id"))]
      # dt[order(data.table:::forder(dt, "id"))] Running twice is probably sufficient to show the caching in action.
    },
    Regression = "c152ced0e5799acee1589910c69c1a2c6586b95d", # Parent of the merge commit that fixes the regression
    Fixed = "1a84514f6d20ff1f9cc614ea9b92ccdee5541506"), # Merge commit of the PR (https://github.com/Rdatatable/data.table/pull/4386/commits) that fixes the regression

Successfully got it to show the improvement using my action in this PR on my fork. (8 hours)

• July 31, ‘24:

1) Continuing work on my slides. Created custom complex objects for the slides which show the Timeline, more/enchanced written content/resources, data.table syntax, and performance testing version labels. Trying to figure different layouts for all the slides. (7 hours)

2) Generated some plots for JSM 2024 and contacted GitHub support to remove an inadvertent PR I made for the same (made to Rdatatable/data.table instead of my fork..wish the default was set to one’s fork honestly) (2 hours)

• Aug 1, ‘24:

• Finished my slides and uploaded them to PresenterHub for the JSM session I’ll be giving a talk in. (7 hours)

• Aug 2, ‘24:

1) Went to Portland, got settled and communicated via email with the rest of the team (Tyson, Kelly, Michael, Jonathan). I shared details about slide re-uploading time limit, format, and rooms (breakout and our session) after communicating with PresenterHub. Reviewed #6344. (8 hours)