Joey, when we talked I said that I thought the effect of specifying a starting commit (be it a week ago, or whenever) was that only commits after that would be "fortified" with submodules, but that earlier commits would still be included in the mirror.

However, that was wrong.  In the mirror repo, the oldest commit is the one corresponding to the one following the specified starting point in the origin.

Does that mean that our idea of creating this new, only-a-week-old experimental mirror repo is useless?

No, I think we should do it anyway. We don't care much about the history for the purposes of this prototype, I think.

I think the main reason to get it up and running is to have a version of chromium/src at HEAD, with submodules, that we can test xrefs on.

The following revision refers to this bug:
  https://chromium.googlesource.com/infra/infra/+/a3b419cb95370f898f8f59aca4ee02adc75afb8f

commit a3b419cb95370f898f8f59aca4ee02adc75afb8f
Author: Alan Bram <flyboy@chromium.org>
Date: Thu Oct 18 00:15:50 2018

Accept starting commit hash as a command-line arg.

Bug: 895967
Change-Id: I2db20d09b78ea18d4928c4908f661d69e11bd507
Reviewed-on: https://chromium-review.googlesource.com/c/1286227
Reviewed-by: Joey Scarr <jsca@chromium.org>
Commit-Queue: Alan Bram <flyboy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#18415}
[modify] https://crrev.com/a3b419cb95370f898f8f59aca4ee02adc75afb8f/infra/services/gsubmodd/gsubmodd.py
[modify] https://crrev.com/a3b419cb95370f898f8f59aca4ee02adc75afb8f/infra/services/gsubmodd/__main__.py
[modify] https://crrev.com/a3b419cb95370f898f8f59aca4ee02adc75afb8f/infra/services/gsubmodd/test/gsubmodd_test_definitions.py
[add] https://crrev.com/a3b419cb95370f898f8f59aca4ee02adc75afb8f/infra/services/gsubmodd/test/gsubmodd_test.expected/epoch_nonexist.json
[modify] https://crrev.com/a3b419cb95370f898f8f59aca4ee02adc75afb8f/infra/services/gsubmodd/test/gsubmodd_test.py
[add] https://crrev.com/a3b419cb95370f898f8f59aca4ee02adc75afb8f/infra/services/gsubmodd/test/gsubmodd_test.expected/epoch.json

The following revision refers to this bug:
  https://chromium.googlesource.com/infra/infra/+/b98462948ebde10ce49ccf2e5e9a732a37302541

commit b98462948ebde10ce49ccf2e5e9a732a37302541
Author: Alan Bram <flyboy@chromium.org>
Date: Thu Oct 18 00:57:27 2018

Plumb through the new "epoch" parameter in the gsubmodd recipe.

Bug: 895967
Change-Id: I2a5cd2dc08caf2820e77d2bf5bd7ce2485ec048f
Reviewed-on: https://chromium-review.googlesource.com/c/1287151
Reviewed-by: Joey Scarr <jsca@chromium.org>
Commit-Queue: Alan Bram <flyboy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#18419}
[modify] https://crrev.com/b98462948ebde10ce49ccf2e5e9a732a37302541/recipes/recipes/gsubmodd.py
[modify] https://crrev.com/b98462948ebde10ce49ccf2e5e9a732a37302541/recipes/README.recipes.md
[add] https://crrev.com/b98462948ebde10ce49ccf2e5e9a732a37302541/recipes/recipes/gsubmodd.expected/epoch.json

The following revision refers to this bug:
  https://chromium.googlesource.com/infra/infra/+/eb01039b772ca16db17ff4dc0daa0216a96e18c3

commit eb01039b772ca16db17ff4dc0daa0216a96e18c3
Author: Alan Bram <flyboy@chromium.org>
Date: Fri Oct 19 00:10:37 2018

Fix bug where new `epoch` arg didn't get plumbed all the way through.

Bug: 895967
Change-Id: I2b8518ea8b0b743a672075924cbaf2b38b206618
Reviewed-on: https://chromium-review.googlesource.com/c/1288999
Reviewed-by: Joey Scarr <jsca@chromium.org>
Commit-Queue: Alan Bram <flyboy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#18465}
[modify] https://crrev.com/eb01039b772ca16db17ff4dc0daa0216a96e18c3/infra/services/gsubmodd/__main__.py