| commit | b4a7cfb38cca2d2f475942b254973ae1d346234f | |
| author | Jingze Wu <jjwu@fb.com> | |
| Wed, 16 Nov 2016 23:31:51 +0000 (16 15:31 -0800) | ||
| committer | Hhvm Bot <hhvm-bot-bot@fb.com> | |
| Wed, 16 Nov 2016 23:37:59 +0000 (16 15:37 -0800) | ||
| tree | b45a35a8c0ad3c55d6b7e04a790e40458d084021 | treesnapshot (tar.gz zip) |
| parent | 381154e90347b35bf1cf5092ac6425f504776a3e | commitdiff |
Added file_with_modes object, store it in saved state
Summary:
In order for lazy parsing to work, we need the file mode of every file, which isn't saved in our fast object. This adds a fast_with_modes object which essentially just stores a file mode with every set of toplevel names.
This doesn't increase the saved state size significantly(< 1MB before compression), and makes it a lot cheaper than reading and parsing modes on the fly.
(Flaky tests, so overriding unit failures)
Reviewed By: andrewjkennedy
Differential Revision: D4085054
fbshipit-source-id: 4a86541c5f504632df2bd031c7342b4cd2bed9f5
Summary:
In order for lazy parsing to work, we need the file mode of every file, which isn't saved in our fast object. This adds a fast_with_modes object which essentially just stores a file mode with every set of toplevel names.
This doesn't increase the saved state size significantly(< 1MB before compression), and makes it a lot cheaper than reading and parsing modes on the fly.
(Flaky tests, so overriding unit failures)
Reviewed By: andrewjkennedy
Differential Revision: D4085054
fbshipit-source-id: 4a86541c5f504632df2bd031c7342b4cd2bed9f5