I seems "subgrid repeat(auto-fill,...)" never worked correctly, but since we disabled subgrid support by default (bug 1221677) I didn't notice.

I think I just misunderstood the code when I created that extra list there. I missed that because serialization then suppressed empty line names lists for subgrid, which we shouldn't do since they are significant for subgrid. (unlike non-subgrid track lists where the track size is the significant part)

Comment on attachment 8711339 [details] [diff] [review] fix - // Array(n + 1).join(s) is a hack for the non-standard s.repeat(n) + // Array(n).join(s) is a hack for the non-standard s.repeat(n) The old comment was actually correct, and the new one isn't. I'll tweak the new comment to say "s.repeat(n - 1)" instead before landing (unless you prefer the old comment). The new expected value is correct though: Array(10000) gives us 9999 [a] plus the leading [foo] sums up to 10000 lines which is the max line number (in an 1-based grid).

Thanks for that clarification. Your planned tweak in comment 3 sounds good.

Comment on attachment 8711339 [details] [diff] [review] fix Review of attachment 8711339 [details] [diff] [review]: ----------------------------------------------------------------- r=me

Comment on attachment 8711340 [details] [diff] [review] enable subgrid for tests Review of attachment 8711340 [details] [diff] [review]: ----------------------------------------------------------------- r=me

https://hg.mozilla.org/integration/mozilla-inbound/rev/b691502e86f8 https://hg.mozilla.org/integration/mozilla-inbound/rev/283da1a207ee

https://hg.mozilla.org/mozilla-central/rev/b691502e86f8 https://hg.mozilla.org/mozilla-central/rev/283da1a207ee

https://hg.mozilla.org/mozilla-central/rev/b691502e86f8 https://hg.mozilla.org/mozilla-central/rev/283da1a207ee