`flex_xvert0` stored the Cartesian position of the vertices in `qpos0`. This array was unused, so it is now replaced with `flex_vert0`, which still contains the positions of the vertices, but normalized on the bounding box of the flex so that the coordinates are in `[0,1]^m->flex_dim`. This will be useful in the future for using different interpolation methods for computing flex deformations.
PiperOrigin-RevId: 691779361
Change-Id: I5c4223103cd4558f4e268fe3e8d8177541e4754f
This allows to resize parent keyframes also when the child has no keyframes.
PiperOrigin-RevId: 691045299
Change-Id: I3ab6509b13b5754e937ec77daf9efe697cabf1bc
The introduction of site specification to connects and welds in 3.2.3 conditionally changed the semantics of `mjData.eq_obj1id` and `mjData.eq_obj2id`. These changes were not properly propagated in several places leading to incorrect computations of constraint inertia, readings of affected force/torque sensors and runtime enabling/disabling of such constraints.
PiperOrigin-RevId: 690670420
Change-Id: I55ee8a013cbee8457f8d6c7f33c2981aedafbab6
Note:
- Move compiler options into `mjsCompiler` struct.
- The compiler options from the attached model are not written by mj_SaveXML.
PiperOrigin-RevId: 689391440
Change-Id: I1d63c146a32f87c737b7a55b64a54b0ffe3aecc9
For `mjs_attachFrame` and `mjs_attachBody`, require either a prefix or suffix.
It is tempting for users to not use a prefix (see #2162), but this causes issues. For example, it creates a duplicate "main" default class in the parent as a child of the already existing "main" default.
PiperOrigin-RevId: 688535755
Change-Id: I653181ca35943f1e4eb89c70fca8b5289818324e
The qposadr_ and dofadr_ variables of the joints were not being reset to -1 when a joint is copied. This could lead to errors when calling mj_recompile after attaching a compiled spec.
PiperOrigin-RevId: 686152004
Change-Id: I9e058089ce87ea0d8e51a7f43475bc48c6527a26
The changes in 9a0dc20821, moving the Hessian from the stack to the arena, should be rolled back: they prevent future threading over islands. Unlike the stack, arena allocations are not thread-friendly. The reason for the original move was to save memory, but the savings are small: `O(ctx->nH)` and only linear in `nv`. The significant reduction of the Cholesky factor, from quadratic in `nv` to quadratic in the largest dof island — the reduction afforded by 2dd518734f — remains in place.
Also refactor and improve readability.
PiperOrigin-RevId: 685717356
Change-Id: Ia9ec3e44a62d459a3b9cffb578cf6479d5aa1d7f
The `extension` section used to discard the plugin name and write the instance name in `plugin.name`, while all of the other sections wrote the instance name in `plugin.instance_name` and optionally the plugin name in `plugin.name` (via the function `OnePlugin`). Now all plugin names are in `plugin.plugin_name` and the instance is written in `plugin.name`, so that it is possible to lookup a plugin via its unique instance name.
PiperOrigin-RevId: 683183804
Change-Id: I179ef5d221143b720db24d745b09e327bf1449ef
This enables to generate a list of all body children of a given type.
Also, add recursive option to mjs_firstChild and mjs_nextChild. Fixes#2112.
PiperOrigin-RevId: 681039348
Change-Id: Iebbb26fd5ec314150206edfdddd88633baa0bec6
- Modify parse_benchmark_test to not time the first instance of file loading.
- Improve readability of step_benchmark_test.
PiperOrigin-RevId: 680953562
Change-Id: I4569d1b946ecb49b6590ea313f0b598ffc897888
Before this CL, the same mjCPlugin objects were referenced by both models.
Fixes#2100.
PiperOrigin-RevId: 680579479
Change-Id: Ida3459f78f9fa4a942cea31b3d5d3e8da2832feb
This is a no-op change for keyframes but it shows an example of behavior.
Also, fixed two bugs:
- Missing namespace when attaching frames.
- Incorrect body id and subtreedof when attaching a compiled spec. Fixes#2063.
PiperOrigin-RevId: 680289467
Change-Id: I2ddaa7ad2226fa466ec3bbf74a7553d497bb4bb9
Taylor Howell