The reftest default preference lists are out of sync between desktop (reftest-cmdline.js) and Android/B2G (bootstrap.js). Two of the diffs are probably prefs that only affect desktop or were for perf reasons (bug 846890, bug 870100), so unlikely to cause bustage if we sync them to bootstrap.js. The other two were introduce in bug 975931 and likely to be substantive. We should fix this code duplication with the preprocessor ASAP, at least once the bug 975931 prefs are sync'd over and green.

Oh, worse, there's a third location in b2g_start_script.js that was added without adding a comment at the other locations.

It turns out syncing the preferences over doesn't introduce any new failures; there are no differences between these two try runs (the failures are due to the patches in the first run): https://tbpl.mozilla.org/?tree=Try&rev=eb9a8b5b0b13 https://tbpl.mozilla.org/?tree=Try&rev=c458891861d7

(And I also have a patch for the preprocessing, but I still want to wait for try results on it.)

Note that the reordering in Makefile.in is because the addition to PP_TARGETS must be before including rules.mk. (Logically, I was moving _DEST_DIR, _HARNESS_FILES, and the _HARNESS_PP_FILES chunk up, but diff shows it the other way around.)

Comment on attachment 8395261 [details] [diff] [review] patch 2 - Sync higher-risk APZC/viewport prefs from bug 975931 from desktop reftest code to Android/B2G reftest code Review of attachment 8395261 [details] [diff] [review]: ----------------------------------------------------------------- Sure would be nice if we could share these prefs in a single location!

Ah you did :-)

https://hg.mozilla.org/integration/mozilla-inbound/rev/0ce172037ced https://hg.mozilla.org/integration/mozilla-inbound/rev/6930ffbc0f1d

https://hg.mozilla.org/mozilla-central/rev/0ce172037ced https://hg.mozilla.org/mozilla-central/rev/6930ffbc0f1d

Comment on attachment 8395302 [details] [diff] [review] patch 3 - Use preprocessor to avoid duplicating reftest default preference lists Review of attachment 8395302 [details] [diff] [review]: ----------------------------------------------------------------- As much as I hate adding more to our Makefiles, this is better than the existing scenario.