James C. (Jamie) Davis

If you like my "metadata vs. data" proposal, then I think it makes sense to refactor `Commit` to contain a `ChangeSet` and a `List`. The data duplicated between ChangeSet and...

Right, at the moment we load all the `ChangeSet`s at once, filter, and then one-at-a-time build `Commit`s. But the *total* cost of the `Commit`s is equal to the overall size...

@mauricioaniche, commits 91f3bee and 78537b3 refactor Commit to use a ChangeSet, and add a new SCM API `getCommit(Changeset)` to avoid duplicating the ChangeSet when creating a Commit from a ChangeSet.

@mauricioaniche Right, my point is just that even if we don't run out of memory, we still don't want to pay the cost of loading every `Commit` if we can...

ee75977 fixes merge conflicts.

Hey @mauricioaniche, had any time for experiments on this?

Paper deadline on Dec 1, I should be able to look at this around Dec. 5?

The semester is almost over... On Sat, Dec 16, 2017 at 5:00 PM Maurício Aniche wrote: > @davisjam See the latest build in the > master and the build at...

I'm a bit torn. I think the additional fields are useful. I just don't need them at the moment. Open to thoughts of other RepoDriller users on reviving this PR...

I see relevant minutes in [2017-08-01/minutes.md](https://github.com/nodejs/benchmarking/blob/master/meetings/2017-08-01/minutes.md), and I guess the issue referenced there is #126. Is #126 still accurate? If so, I suggest we put that information somewhere reasonably prominently...