only call scalar finite when necessary

only call scalar finite when necessary

Mike points that the the ulps compares
rank high in path ops profiles. The
check for finite scalars is rarely
required.

Call it less by:
- specializing _pin version of compares
- checking for 0 divides up front
- handling failing cases before comparing
- casting float to double before adding

R=reed@google.com

Committed: https://skia.googlesource.com/skia/+/b669300a9753893ef900207c38aeff2d467764e5

[caryclark, 2015-12-14 20:47:14]

The CQ bit was checked by caryclark@google.com to run a CQ dry run

[commit-bot: I haz the power, 2015-12-14 20:47:17]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1522183002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1522183002/1

[commit-bot: I haz the power, 2015-12-14 21:05:59]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-12-14 21:06:00]

Dry run: This issue passed the CQ dry run.

[caryclark, 2015-12-15 13:15:19]

PTAL

[reed1, 2015-12-15 14:17:21]

lgtm

https://codereview.chromium.org/1522183002/diff/1/src/pathops/SkPathOpsTypes.cpp
File src/pathops/SkPathOpsTypes.cpp (right):

https://codereview.chromium.org/1522183002/diff/1/src/pathops/SkPathOpsTypes....
src/pathops/SkPathOpsTypes.cpp:28: static bool equal_ulps_pin(float a, float b,
int epsilon, int depsilon) {
Just an idea, may not be any clearer...

To avoid the copy/paste here and in no_equal_ulps[_pin], we could use a template
arg...

template <bool CheckIsFinite> equal_ulps(float a, float b, ...) {
    if (CheckIsFinite) {
        if (...) {
        }
    }
    ...

https://codereview.chromium.org/1522183002/diff/1/src/pathops/SkPathOpsTypes.h
File src/pathops/SkPathOpsTypes.h (right):

https://codereview.chromium.org/1522183002/diff/1/src/pathops/SkPathOpsTypes....
src/pathops/SkPathOpsTypes.h:170: bool AlmostEqualUlps_Pin(float a, float b);
Pin didn't convey anything obvious to me when I saw this and its call-sites.

Not sure I have a better word (other than the ugly _CheckForFinite), so perhaps
we just need some Dox here in the header.

[caryclark, 2015-12-16 20:13:23]

The CQ bit was checked by caryclark@google.com

[commit-bot: I haz the power, 2015-12-16 20:13:27]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1522183002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1522183002/1

[commit-bot: I haz the power, 2015-12-16 20:28:37]

Description was changed from

==========
only call scalar finite when necessary

Mike points that the the ulps compares
rank high in path ops profiles. The
check for finite scalars is rarely
required.

Call it less by:
- specializing _pin version of compares
- checking for 0 divides up front
- handling failing cases before comparing
- casting float to double before adding

R=reed@google.com
==========

to

==========
only call scalar finite when necessary

Mike points that the the ulps compares
rank high in path ops profiles. The
check for finite scalars is rarely
required.

Call it less by:
- specializing _pin version of compares
- checking for 0 divides up front
- handling failing cases before comparing
- casting float to double before adding

R=reed@google.com

Committed:
https://skia.googlesource.com/skia/+/b669300a9753893ef900207c38aeff2d467764e5
==========

[commit-bot: I haz the power, 2015-12-16 20:28:38]

Committed patchset #1 (id:1) as
https://skia.googlesource.com/skia/+/b669300a9753893ef900207c38aeff2d467764e5