Refactor: make evaluate_fcurve() take its FCurve as const instead of mutable #121788

Nathan Vegdahl · 2024-05-14T15:41:45+02:00

Nathan Vegdahl commented

2024-05-14 15:41:45 +02:00

This required making a whole bunch of other functions in the call chain
take const parameters as well. It also required changing some function
pointers in some types to take const parameters, which in turn required
changing all the functions that are pointed to by those function
pointers to take const parameters as well.

Additionally, there was one mutable usage of the FModifier * parameter
in fcm_cycles_time() that had to be removed to make the call chain
const. However, this turned out to be a code path that shouldn't be
reachable, and would represent a bug elsewhere. So it was changed to
an assert.

All in all, the non-constness was deep and tangled.

There's still a lot more that we can make const, but I wanted to keep
this change as narrow and focused as possible.

Notes for Reviewers

There are other parameters that can trivially be made const at the function definitions (like floats, ints, etc.). I've intentionally left those alone, to focus this on the minimal set of changes needed to make evaluate_fcurve()'s FCurve * parameter const.

This required making a whole bunch of other functions in the call chain take const parameters as well. It also required changing some function pointers in some types to take const parameters, which in turn required changing all the functions that are pointed to by those function pointers to take const parameters as well. Additionally, there was one mutable usage of the `FModifier *` parameter in `fcm_cycles_time()` that had to be removed to make the call chain const. However, this turned out to be a code path that shouldn't be reachable, and would represent a bug elsewhere. So it was changed to an assert. All in all, the non-constness was deep and tangled. There's still a lot more that we can make const, but I wanted to keep this change as narrow and focused as possible. ## Notes for Reviewers - There are other parameters that can trivially be made const at the function definitions (like floats, ints, etc.). I've intentionally left those alone, to focus this on the minimal set of changes needed to make `evaluate_fcurve()`'s `FCurve *` parameter const.

Nathan Vegdahl added the

Module

Animation & Rigging

label 2024-05-14 15:41:45 +02:00

Nathan Vegdahl added 1 commit 2024-05-14 15:41:54 +02:00

Make evaluate_fcurve() take its FCurve as const instead of mutable faa72df387

This required making a whole bunch of other functions in the call chain
take const parameters as well.  It also required changing some function
pointers in some types to take const parameters, which in turn required
changing all the functions that are pointed to by those function
pointers to take const parameters as well.

All in all, the non-constness was deep and tangled.

There's still a lot more that we can make const, but I wanted to keep
this change as narrow and focused as possible.

Nathan Vegdahl reviewed 2024-05-14 15:43:00 +02:00

source/blender/blenkernel/intern/fmodifier.cc Outdated

						
				@ -627,3 +636,3 @@

				  /* FIXME... */

				  if (fcm->prev) {

				    fcm->flag |= FMODIFIER_FLAG_DISABLED;

				    // fcm->flag |= FMODIFIER_FLAG_DISABLED;

Nathan Vegdahl commented

2024-05-14 15:43:00 +02:00

Here is the (likely bogus) change that's otherwise blocking evaluate_fcurve() from being const.

Here is the (likely bogus) change that's otherwise blocking `evaluate_fcurve()` from being const.

Falk David commented

2024-05-14 16:14:11 +02:00

Looks like this logic would have to be pulled out and put into evaluate_time_fmodifiers directly (that's where this code likely should have been from the start).

Looks like this logic would have to be pulled out and put into `evaluate_time_fmodifiers` directly (that's where this code likely should have been from the start).

Falk David commented

2024-05-14 16:20:00 +02:00

diff --git a/source/blender/blenkernel/intern/fmodifier.cc b/source/blender/blenkernel/intern/fmodifier.cc
index 025c4851733..7e49d71edee 100644
--- a/source/blender/blenkernel/intern/fmodifier.cc
+++ b/source/blender/blenkernel/intern/fmodifier.cc
@@ -623,13 +623,6 @@ static float fcm_cycles_time(
   /* Initialize storage. */
   storage->cycyofs = 0;
 
-  /* check if modifier is first in stack, otherwise disable ourself... */
-  /* FIXME... */
-  if (fcm->prev) {
-    fcm->flag |= FMODIFIER_FLAG_DISABLED;
-    return evaltime;
-  }
-
   if (fcu == nullptr || (fcu->bezt == nullptr && fcu->fpt == nullptr)) {
     return evaltime;
   }
@@ -1399,6 +1392,19 @@ float evaluate_time_fmodifiers(FModifiersStackStorage *storage,
     return evaltime;
   }
 
+  /**
+   * Disable all cycles modifiers that are not first in the stack.
+   */
+  LISTBASE_FOREACH (FModifier *, fcm, modifiers) {
+    const FModifierTypeInfo *fmi = fmodifier_get_typeinfo(fcm);
+    if (fmi == nullptr || fmi->type != FMODIFIER_TYPE_CYCLES) {
+      continue;
+    }
+    if (fcm->prev) {
+      fcm->flag |= FMODIFIER_FLAG_DISABLED;
+    }
+  }
+
   /* Starting from the end of the stack, calculate the time effects of various stacked modifiers
    * on the time the F-Curve should be evaluated at.
    *

```diff diff --git a/source/blender/blenkernel/intern/fmodifier.cc b/source/blender/blenkernel/intern/fmodifier.cc index 025c4851733..7e49d71edee 100644 --- a/source/blender/blenkernel/intern/fmodifier.cc +++ b/source/blender/blenkernel/intern/fmodifier.cc @@ -623,13 +623,6 @@ static float fcm_cycles_time( /* Initialize storage. */ storage->cycyofs = 0; - /* check if modifier is first in stack, otherwise disable ourself... */ - /* FIXME... */ - if (fcm->prev) { - fcm->flag |= FMODIFIER_FLAG_DISABLED; - return evaltime; - } - if (fcu == nullptr || (fcu->bezt == nullptr && fcu->fpt == nullptr)) { return evaltime; } @@ -1399,6 +1392,19 @@ float evaluate_time_fmodifiers(FModifiersStackStorage *storage, return evaltime; } + /** + * Disable all cycles modifiers that are not first in the stack. + */ + LISTBASE_FOREACH (FModifier *, fcm, modifiers) { + const FModifierTypeInfo *fmi = fmodifier_get_typeinfo(fcm); + if (fmi == nullptr || fmi->type != FMODIFIER_TYPE_CYCLES) { + continue; + } + if (fcm->prev) { + fcm->flag |= FMODIFIER_FLAG_DISABLED; + } + } + /* Starting from the end of the stack, calculate the time effects of various stacked modifiers * on the time the F-Curve should be evaluated at. * ```

Nathan Vegdahl commented

2024-05-14 16:30:15 +02:00

Thanks! That may indeed be the case.

However, that doesn't solve the problem because evaluate_time_fmodifiers() also needs to take those parameters as const for this PR. So it might be a good refactor, but isn't what's needed for this PR.

Thanks! That may indeed be the case. However, that doesn't solve the problem because `evaluate_time_fmodifiers()` *also* needs to take those parameters as const for this PR. So it might be a good refactor, but isn't what's needed for this PR.

Falk David commented

2024-05-14 17:50:51 +02:00

Ah you're right indeed.
Looking at add_fmodifier, it seems that it's already not possible to add a FMODIFIER_TYPE_CYCLES after the first one. Maybe this just needs to be an assert instead? Is that code actually ever hit? 😅

Ah you're right indeed. Looking at `add_fmodifier`, it seems that it's already not possible to add a `FMODIFIER_TYPE_CYCLES` after the first one. Maybe this just needs to be an assert instead? Is that code actually ever hit? 😅

Christoph Lendenfeld commented

2024-05-16 12:10:20 +02:00

I agree with @filedescriptor here. I don't think you can have a cycles modifier at any other place than the first one. Changing that out with an assert is a valid option. And in the long run we can maybe remove that limitation altogether.

👍 1

Falk David commented

2024-05-16 12:16:21 +02:00

Yes, also from what I can tell, all codepaths that add an fcurve modifier go through add_fmodifier. Even the Python API uses it (see rna_FCurve_modifiers_new).

Yes, also from what I can tell, all codepaths that add an fcurve modifier go through `add_fmodifier`. Even the Python API uses it (see `rna_FCurve_modifiers_new`).

👍 1

nathanvegdahl marked this conversation as resolved

Nathan Vegdahl requested review from Sybren A. Stüvel 2024-05-14 15:43:40 +02:00

Nathan Vegdahl requested review from Christoph Lendenfeld 2024-05-14 15:43:49 +02:00

Nathan Vegdahl changed title from ~~Make `evaluate_fcurve()` take its FCurve as const instead of mutable~~ to WIP: Make `evaluate_fcurve()` take its FCurve as const instead of mutable

2024-05-14 15:51:11 +02:00

Nathan Vegdahl referenced this pull request

2024-05-14 15:53:58 +02:00

Anim: implement 3D viewport keyframing functionality for the Animation data-block #120428

Nathan Vegdahl commented

2024-05-14 15:57:05 +02:00

Context: this PR was inspired by a comment from @dr.sybren on #120428 (now #121661) about making new_key_needed() take a const FCurve reference.

Context: this PR was inspired by a comment from @dr.sybren on #120428 (now #121661) about making `new_key_needed()` take a const FCurve reference.

Nathan Vegdahl added 2 commits 2024-05-16 15:22:10 +02:00

Merge branch 'main' into const_fcurve_functions aa5637d696

Change cycle-modifier-is-first check into an assert

buildbot/vexp-code-patch-lint Build done.

Details

buildbot/vexp-code-patch-linux-x86_64 Build done.

Details

buildbot/vexp-code-patch-darwin-x86_64 Build done.

Details

buildbot/vexp-code-patch-darwin-arm64 Build done.

Details

buildbot/vexp-code-patch-windows-amd64 Build done.

Details

buildbot/vexp-code-patch-coordinator Build done.

Details

fe7539e304

Nathan Vegdahl changed title from ~~WIP: Make `evaluate_fcurve()` take its FCurve as const instead of mutable~~ to Make `evaluate_fcurve()` take its FCurve as const instead of mutable

2024-05-16 15:27:46 +02:00

Nathan Vegdahl requested review from Falk David 2024-05-16 15:28:04 +02:00

Nathan Vegdahl commented

2024-05-16 15:28:52 +02:00

@ChrisLend @filedescriptor I turned that one mutable usage into an assert as suggested. I think this is ready to land, pending a final review.

Nathan Vegdahl commented

2024-05-16 15:29:04 +02:00

@blender-bot build

Christoph Lendenfeld approved these changes 2024-05-16 16:40:01 +02:00

Christoph Lendenfeld left a comment

The title should probably be Refactor:... but else the PR looks good.
Thanks for that, much nicer!

The title should probably be `Refactor:...` but else the PR looks good. Thanks for that, much nicer!

👍 1

Nathan Vegdahl changed title from ~~Make `evaluate_fcurve()` take its FCurve as const instead of mutable~~ to Refactor: make `evaluate_fcurve()` take its FCurve as const instead of mutable

2024-05-16 17:10:45 +02:00

Nathan Vegdahl merged commit 848bd505b4 into main

2024-05-16 17:13:53 +02:00

Nathan Vegdahl deleted branch const_fcurve_functions

2024-05-16 17:13:56 +02:00

Nathan Vegdahl referenced this pull request

2024-05-16 17:47:34 +02:00

Refactor: make yet more fcurve evaluation functions take const fcurves #121882

Nathan Vegdahl referenced this issue from a commit

2024-05-17 15:57:07 +02:00

Refactor: make yet more fcurve evaluation functions take const fcurves

Sign in to join this conversation.