Details

Reviewers

Summary

Use threading::parallel_for for multi-threading, for a simpler API, more readable and concise code.
Use Span and Array (only internally, the public API is still C) for safer, more automatic memory management.
Since code is much less verbose, combine the callbacks into the main function. Note that the accumulation code could be more concise with float3, I just wanted to keep these changes minimal.

@Campbell Barton (campbellbarton), do you have any of the test files you used from D11993?
I'd check those to make sure this doesn't have a performance impact (I wouldn't expect it to).

Diff Detail

Repository

rB Blender

Branch

cleanup-normal-calculation (branched from master)

Build Status

Buildable 19370
Build 19370: arc lint + arc unit

Event Timeline

Hans Goudey (HooglyBoogly) requested review of this revision.Sep 4 2021, 9:58 PM

Hans Goudey (HooglyBoogly) created this revision.

Harbormaster completed remote builds in B16818: Diff 41466.Sep 4 2021, 9:58 PM

Hans Goudey (HooglyBoogly) edited the summary of this revision. (Show Details)Sep 4 2021, 9:59 PM

Hans Goudey (HooglyBoogly) added a reviewer: Campbell Barton (campbellbarton).

Hans Goudey (HooglyBoogly) added a project: Modeling.

Harbormaster completed remote builds in B17609: Diff 42987.Oct 6 2021, 8:20 PM

Used incorrect diff before

Hans Goudey (HooglyBoogly) edited the summary of this revision. (Show Details)Oct 6 2021, 8:23 PM

Hans Goudey (HooglyBoogly) mentioned this in D12770: Refactor: Move normals out of MVert, lazy calculation.Nov 26 2021, 7:18 PM

Merge branch 'master' into temp-vert-normals-cleanup

Harbormaster completed remote builds in B19143: Diff 45561.Dec 2 2021, 4:46 AM

Merge branch 'master' into temp-vert-normals-cleanup
Fix: Incorrect assert

I'm currently observing very weird assert behavior in Erindale's flower shop file:

BLI_assert(CustomData_has_layer(&mesh->vdata, CD_NORMAL) || mesh->totvert == 0);

(gdb) print (CustomData_has_layer(&mesh->vdata, CD_NORMAL) || mesh->totvert == 0)
$6 = true

(gdb) print mesh->vdata.layers[1].type
$8 = 8  # CD_NORMAL = 8

These bits of information just don't seem to line up!
However, running Blender with -t 1 makes the issue go away.
Maybe another thread modified the mesh right after this one aborted.

Harbormaster completed remote builds in B19341: Diff 45889.Dec 9 2021, 11:56 PM