[19:24:01] chad@notsexy /a/mw-pristine (REL1_29)$ git log --no-merges origin/REL1_28..origin/REL1_29 | wc -l [19:24:01] 20188 [19:24:05] That can't possibly be right. [19:24:12] 20k commits this release? [19:25:04] Oh, 1640 [19:25:07] Forgot --oneline [19:25:08] dur [19:52:24] RainbowSprinkles: xD [19:53:54] git rev-list --count also does what I want [19:53:55] But w/e [19:53:56] That's only 12 lines commit msg on average? (Presumably 9, given -3 for hash date and separating blank line) [19:53:57] Done now [19:54:24] Does wc count blank lines as lines? [19:54:39] I judge the quality of a release by the average size of its commit messages [19:54:43] commit 4a3fb7811c8e7e5a1801e9f4583cf1c485a947c0 [19:54:43] Author: Chad Horohoe [19:54:43] Date: Thu Jul 13 08:56:21 2017 -0700 [19:54:43] Bump version number for 1.29.0 [19:54:43] [19:54:43] Change-Id: Iac58039fd4624ccf9b44baf359d6304264494ea7 [19:54:58] So like is that 5 or 7? [19:55:22] Krinkle: The best commits change one line and have 4-5 paragraphs explaining why :) [19:55:35] Yup [19:55:42] At least we know the inverse is true [19:56:04] Big commits with a one line summary [19:56:09] Always ends badly [19:57:20] Sounds like a good starting point for one of TrevorParscal's http://www.theoreticallogic.com/ [19:58:13] Heh [21:17:04] Krinkle: l10n-bot commits and merge commits are one-liners [21:17:29] well, five-liners and for-liners, respectively [21:17:38] but they pull the average down