(In reply to David Major [:dmajor] from comment #7) > Before: > ``` > NSS section sizes opt nightly: 2370010 > XUL section sizes opt nightly: 94344820 > ``` > > After: > ``` > NSS section sizes opt nightly: 2618210 > XUL section sizes opt nightly: 95536532 > ``` > > So libxul gained more than a megabyte, and NSS gained a pretty alarming 10%. Was the difference in these pushes just PGO settings? Or did some feature code land in between or something? Looks like yes: Empty trypush (https://treeherder.mozilla.org/#/jobs?repo=try&revision=f1e6f9b8d9663f092e0f27a3c88e49aec2646536&selectedJob=254434385 ): > installer size opt nightly: 71145736 > NSS section sizes opt nightly: 2630178 > XUL section sizes opt nightly: 96357452 Backout (so e10s turned off; https://treeherder.mozilla.org/#/jobs?repo=try&revision=83f72cd9f01390fc93eb72bbf03f10b47d3097a5&selectedJob=254433998 ): > installer size opt nightly: 70097095 > NSS section sizes opt nightly: 2360946 > XUL section sizes opt nightly: 94241652 This isn't really my area of expertise - do we have a way of inspecting/diffing the resulting binaries and finding out what the cause of the difference is? I'll note that there was a similar increase on the 64-bit builds, which was somehow not flagged up in the same way, which comes back to my earlier questions of why only win7 complained here...
Bug 1559878 Comment 9 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
(In reply to David Major [:dmajor] from comment #7) > Before: > ``` > NSS section sizes opt nightly: 2370010 > XUL section sizes opt nightly: 94344820 > ``` > > After: > ``` > NSS section sizes opt nightly: 2618210 > XUL section sizes opt nightly: 95536532 > ``` > > So libxul gained more than a megabyte, and NSS gained a pretty alarming 10%. Was the difference in these pushes just PGO settings? Or did some feature code land in between or something? Looks like it's just the PGO settings: Empty trypush (https://treeherder.mozilla.org/#/jobs?repo=try&revision=f1e6f9b8d9663f092e0f27a3c88e49aec2646536&selectedJob=254434385 ): > installer size opt nightly: 71145736 > NSS section sizes opt nightly: 2630178 > XUL section sizes opt nightly: 96357452 Backout (so e10s turned off; https://treeherder.mozilla.org/#/jobs?repo=try&revision=83f72cd9f01390fc93eb72bbf03f10b47d3097a5&selectedJob=254433998 ): > installer size opt nightly: 70097095 > NSS section sizes opt nightly: 2360946 > XUL section sizes opt nightly: 94241652 This isn't really my area of expertise - do we have a way of inspecting/diffing the resulting binaries and finding out what the cause of the difference is? I'll note that there was a similar increase on the 64-bit builds, which was somehow not flagged up in the same way, which comes back to my earlier questions of why only win7 complained here...