Updates relating to changes to deploy-ng

Author: rdb

distribution/building-binaries.rst

@@ -176,11 +176,14 @@ combination is:
    * - p3fmod_audio
      - Audio (including 3D audio) support using FMOD (note the licensing!)
    * - pandaegg
-     - Enables support for reading .egg files (usually not necessary)
+     - Enables support for reading .egg files. Generally, you should not include
+       this, since .egg files are automatically converted to .bam during build.
    * - p3ptloader
-     - Adds support for additional model formats beyond BAM
+     - Adds support for additional model formats. You probably want to instead
+       add those model extensions to the ``bam-model-extensions`` list.
    * - p3assimp
-     - Adds support for additional model formats beyond BAM by using Assimp
+     - Adds support for additional model formats. You probably want to instead
+       add those model extensions to the ``bam-model-extensions`` list.
 
 Note that some plug-ins use third-party libraries that may have different
 licensing terms from Panda3D. More information about these libraries can be
@@ -198,17 +201,17 @@ versions. The default set is as follows:
 
 .. code-block:: python
 
-   'platforms': ['manylinux2010_x86_64', 'macosx_10_9_x86_64', 'win_amd64'],
+   'platforms': ['manylinux2014_x86_64', 'macosx_10_9_x86_64', 'win_amd64'],
 
 Sometimes, it is desirable to use third-party packages that do not provide
 wheels for a given platform. For example, a package may no longer publish wheels
-for a ``manylinux2010_x86_64``, but only for ``manylinux2014_x86_64``. If you
-wish to use this package, then you need to therefore set the platform tags to
+for a ``macosx_10_9_x86_64``, but only for ``macosx_10_13_x86_64``. If you wish
+to use this package, then you need to therefore set the platform tags to
 increase these versions:
 
 .. code-block:: python
 
-   'platforms': ['manylinux2014_x86_64', 'macosx_10_9_x86_64', 'win_amd64'],
+   'platforms': ['manylinux2014_x86_64', 'macosx_10_13_x86_64', 'win_amd64'],
 
 .. list-table:: List of Platforms
    :widths: 20, 80
@@ -217,10 +220,14 @@ increase these versions:
      - 64-bit Windows systems (including Intel x64 processors).
    * - win32
      - 32-bit Windows systems, rarely used nowadays.
-   * - manylinux2010_x86_64
-     - Target 64-bit Linux distributions more recent than (more or less) 2010.
-   * - manylinux2010_i686
-     - Target 32-bit Linux distributions more recent than (more or less) 2010.
+   * - manylinux2014_x86_64
+     - Target 64-bit Linux distributions more recent than (more or less) 2014.
+   * - manylinux2014_i686
+     - Target 32-bit Linux distributions more recent than (more or less) 2014.
+   * - linux_x86_64
+     - Target 64-bit Linux distributions, letting Panda decide which version.
+   * - linux_i686
+     - Target 32-bit Linux distributions, letting Panda decide which version.
    * - macosx_10_9_x86_64
      - Target Intel Macs running OS X Mavericks or higher. Recommended.
    * - android

distribution/list-of-build-options.rst

@@ -89,8 +89,23 @@ file_handlers
 
    The function is run when encountering a file with the given extension.
    User-defined file handlers for an extension override the default handler.
-   By default, there is only one file handler registered: for .egg files, which
-   runs egg2bam.
+
+bam_model_extensions
+   A list of model extensions that are automatically converted to .bam during
+   build. By default, this includes .egg, .gltf and .glb. Change this if you
+   wish to use :ref:`other file formats <supported-model-formats>` than these,
+   or if you don't want these models to be converted to .bam during build.
+
+strip_docstrings
+   If true, which is the default, all docstrings will be removed as an
+   optimization, and any use of ``__doc__`` will return ``None``. Set this to
+   false if you need to keep these docstrings for some reason.
+
+prefer_discrete_gpu
+   On systems with both an integrated and dedicated GPUs, tells the driver that
+   the application prefers to use the dedicated GPU, which usually provides
+   higher performance. At the moment, this option is only implemented on
+   Windows, and only for NVIDIA and AMD graphics cards.
 
 application_id
    This field is required by Android and uniquely identifies the application.

distribution/packaging-binaries.rst

@@ -17,8 +17,7 @@ The ``bdist_apps`` command will make use of ``build_apps.build_base`` and
 installers
    A dictionary of installers/packages to create for each platform (keys are
    strings for the platform, values are strings representing installers,
-   defaults to gztar for manylinux2010_x86_64/manylinux2010_i386 and zip for
-   everything else)
+   defaults to gztar for manylinux and zip for everything else)
 
 Installers
 ----------

distribution/setuptools-examples.rst

@@ -39,7 +39,7 @@ Then, we can build the binaries using ``python setup.py build_apps``.
 
 A ``build`` directory will be created and contain a directory for each platform
 that binaries were built for. Since no platforms were specified, the defaults
-were used (manylinux2010_x86_64, macosx_10_9_x86_64, win_amd64).
+were used (manylinux2014_x86_64, macosx_10_9_x86_64, win_amd64).
 
 Note, win32 is missing from the defaults. If a win32 build is desired, then
 platforms must be defined in ``setup.py`` and ``win_amd64`` added to the list:
@@ -54,7 +54,7 @@ platforms must be defined in ``setup.py`` and ``win_amd64`` added to the list:
            'build_apps': {
                'console_apps': {'hello_world': 'main.py'},
                'platforms': [
-                   'manylinux2010_x86_64',
+                   'manylinux2014_x86_64',
                    'macosx_10_9_x86_64',
                    'win_amd64',
                    'win32',

optimization/performance-issues/motherboard-integrated-video.rst

@@ -30,3 +30,49 @@ The following code can be used to determine which GPU is in use:
 
       std::cerr << win->get_gsg()->get_driver_vendor() << "\n"
                 << win->get_gsg()->get_driver_renderer() << "\n";
+
+Forcing Use of Discrete GPU on Windows
+--------------------------------------
+
+On Windows, it is possible to force the NVIDIA and AMD graphics drivers to
+automatically select the high-peformance dedicated graphics card, by compiling
+special symbols into the main executable.
+
+.. only:: python
+
+   While developing, the main executable is ``python.exe``, which does not set
+   these symbols. But when :ref:`building an application <distribution>`,
+   build_apps can automatically add these special symbols. Simply add this
+   option to the build_apps options block in setup.py:
+
+   .. code-block:: python
+
+      'prefer_discrete_gpu': True,
+
+.. only:: cpp
+
+   Simply copy-paste the following symbols into the source file containing your
+   main entry point:
+
+   .. code-block:: cpp
+
+      #ifdef _WIN32
+      extern "C" {
+        __declspec(dllexport) DWORD AmdPowerXpressRequestHighPerformance = 0x00000001;
+        __declspec(dllexport) DWORD NvOptimusEnablement = 0x00000001;
+      }
+      #endif
+
+Forcing Use of Discrete GPU on Linux
+------------------------------------
+
+On Linux, some drivers can be told to use the discrete GPU by setting
+``DRI_PRIME=1`` in the environment. However, this is not considered reliable at
+this time, so it is not done by Panda3D automatically. It is suggested to
+document this as a possibility for your end-users or add an option for this
+setting that can be disabled.
+
+When distributing a .desktop file, it is also possible to add the following key
+to the file::
+
+   PrefersNonDefaultGPU=true

pipeline/gltf-files.rst

@@ -131,6 +131,26 @@ the vertex shader:
 
    binormal = cross(p3d_Normal, p3d_Tangent.xyz) * p3d_Tangent.w
 
+Distributing glTF Models
+------------------------
+
+.. only:: python
+
+   When :ref:`building your application <distribution>`, glTF models with the
+   .gltf or .glb models are converted to .bam by default, assuming that they are
+   referenced in the include_patterns list in setup.py. The .bam extension is
+   appended after the existing extension, like ``model.gltf.bam``. This will
+   allow the model loader to automatically look for ``model.gltf.bam`` when
+   trying to load ``model.gltf``, so that you do not need to change your calls
+   to :py:meth:`loader.loadModel() <direct.showbase.Loader.Loader.loadModel>`.
+
+.. only:: cpp
+
+   To avoid having to include the plug-ins with a distributed application, use
+   the gltf2bam utility (provided with panda3d-gltf) or write a script to
+   convert the model to .bam using :meth:`.NodePath.write_bam_file()` and ship
+   the converted .bam file instead of the original source file.
+
 External Links
 --------------
 

pipeline/model-file-formats.rst

@@ -205,7 +205,8 @@ For several formats for which Panda3D ships with a to-egg conversion tool,
 Panda3D can automatically do the step of converting the model to .egg on load.
 For example, Panda3D ships with a flt2egg converter, which can convert
 OpenFlight models to the Egg format. If you try to load a .flt file, Panda3D
-will implicitly invoke flt2egg behind the scenes.
+will implicitly invoke flt2egg behind the scenes. The plug-in responsible for
+this is called p3ptloader.
 
 The formats supported by this plug-in are OpenFlight (.flt), LightWave (.lwo),
 AutoCAD (.dxf), VRML (.wrl), Direct X (.x), and Wavefront OBJ (.obj).