Persistence of Vision Ray Tracer ('POV-Ray') version 3.8. Copyright 1991-2017 Persistence of Vision Raytracer Pty. Ltd.
POV-Ray is free software: you can redistribute it and/or modify it under the terms of the GNU Affero General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
POV-Ray is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License along with this program. If not, see http://www.gnu.org/licenses/
This document does not attempt to explain how POV-Ray works or how to make modifications to it. Some of the internals of POV-Ray are highly complex; if you wish to gain a better understanding of that topic, please follow the link provided on the source code page mentioned below. This document is intended only to assist you in getting the POV-Ray source code built using an officially supported compiler on Win32 or Win64. Furthermore, no attempt is made to cover advanced subjects such as profile-guided optimization (though we will point out that the latter is definitely worthwhile, if you have the time and patience).
Please use our support groups at http://news.povray.org/ or news://news.povray.org/. These groups provide the best means for support with relation to this subject.
Currently the only officially supported means of building the Windows version of POV-Ray is via the provided Visual Studio 2015 solution.
You will need Visual Studio 2015 (Community Edition will suffice) or later.
Due to the free availability of Visual Studio 2015 Community Edition and better performance of the generated binaries, we have ceased support for earlier versions of Visual Studio.
It is not necessary to build the editor support DLL (cmedit), as it is
loaded via LoadLibrary()
and thus no import library is required. You
can just use the DLL's included with the official binary
distribution. If however you do wish to build it, either enable it in
the configuration manager or initiate a build of that specific project
manually.
cmedit acts as an interface between POVWIN and the full editor DLL (povcmax), which is supplied as a binary with the official POV-Ray for Windows distribution. The source code for povcmax is based on the codemax code editor and is not part of the AGPL3 POV-Ray source distribution. If you wish to know more about the editors, please refer to the "about the editor DLL's" section below.
Note: The editor DLL's are not required for POVWIN to work. If they are not present or cannot be loaded, POVWIN is designed to be able to function quite happily without them. The editor and its related functions will simply not be present.
-
Make sure you have a working copy of the appropriate release binary installed. This ensures that the appropriate registry settings and support files are present. If you don't do this, your compiled code will probably not work on your machine.
-
Open
windows\vs2015\povray.sln
in Visual Studio. Set 'Windows Targets > GUI' as the start-up project if it is not already selected. Then, select the 'Generic POV-Ray > povbase' project and expand 'Backend Headers', then open the filebuild.h
listed within it. Please setBUILT_BY
to your real name (and contact info) to make unofficial versions distinguishable from each other. Remove the#error
directive afterwards to proceed with the build. Note that if you do not do this you will get compile errors.Note: If you are building in a Git local workspace, we recommend you run
git update-index --skip-workspace source/base/build.h
. This will keep your changes to the file from being included in a commit, or overwritten in a pull or reset, while still allowing you to pull non-conflicting updates to the file).By default, the GUI project is built, and the Console project is excluded. If you wish to generate a simple console version of POV-Ray, please modify the configuration as needed, edit
windows/povconfig/syspovconfig.h
to enable the definition of_CONSOLE
as noted therein, and build.Note: The windows console project is intended to be a simple example of how to use the VFE library to make a console version of POV-Ray with minimal code. It is not intended to be a fully-featured console build of POV. If you wish to make a more comprehensive console version that runs on windows, please use the unix version as a guide - it uses the same principles (as it is also linked with VFE) but has more features.
-
Once you have taken whatever steps are needed as set out above, select your desired build (e.g.
Release|Win32
), hit F7, and wait. We recommend that you read the rest of this document while you are waiting.If all goes well, you should end up with the POV-Ray for Windows executable. All 32-bit binaries should end up in
windows\vs2015\bin32
, and the 64-bit ones should be inwindows\vs2015\bin64
.Note: if you are building a DEBUG version of POV-Ray, be aware that the executable will attempt to load the debug version of POV-Ray's editor DLLs. In this case, if you want the editor to be available in the GUI, copy the standard editor DLLs that come with the official POV-Ray distribution to their debug equivalent names (e.g. copy
cmedit32.dll
tocmedit32d.dll
, etc).Note that the copied
cmedit32.dll
will still attempt to loadpovcmax32.dll
(i.e. the non-debug version), so it's not necessary to copypovcmax32.dll
to its debug equivalent. The same goes forpovcmax64.dll
.
Note: You should not attempt to build any of the following targets (if they are present), as they are not actively maintained, and may generate flawed binaries or fail to compile at all:
Release-AVX|Win32
(This target would be useless, as to the best of our knowledge there are no 32-bit processors or operating systems with AVX support.)Release-SSE2|x64
(This target would be redundant, as SSE2 is a standard feature of the x64 architecture and is therefore implicitly enabled in allx64
builds, including vanillaRelease|x64
.)
The only reason for their presence is that Visual Studio keeps insisting on having all possible combinations of targets and configurations, no matter how hard we try to get rid of them.
Please read the rest of this document while you're waiting for your compile to finish, particularly the 'other things you may want to keep in mind' at the end.
By default, binaries generated with Visual Studio 2015 will only run on Windows Vista or later. Visual Studio 2015 can be used to generate binaries compatible with Windows versions as early as Windows XP (SP3 required), but the procedure is a bit tricky, and the following description is therefore aimed at people with in-depth knowledge of Visual Studio and its toolchain.
You will need Visual Studio 2015 (again Community Edition will suffice) Update 1 or later, with the following optional features installed:
- Programming Languages / Visual C++ / Windows XP Support for C++
- Windows and Web Development / Universal Windows App Development Tools / Windows 10 SDK (10.0.10586)
(Later Windows 10 SDKs may also work, but will require you to modify the settings shown below accordingly.)
To enable XP compatibility, you will have to either edit or override the following properties in all Visual Studio project files:
PlatformToolset
must be set tov140_xp
.TargetUniversalCRTVersion
must be set to10.0.10568.0
.
The latter is the tricky part, as the property cannot be accessed via
the IDE, and while there are multiple approaches to achieve this (e.g.
via a user property file, or by manually editing the default in
%ProgramFiles(x86)%\MSBuild\Microsoft.Cpp\v4.0\V140\Microsoft.Cpp.Common.props
)
we highly recommend running the build from the command line using the
msbuild.exe
command-line tool, making use of its /p:
command-line
option to override the properties in question.
For details on the use of the command-line build tool, please consult Microsoft's official documentation.
Attention: Failure to override the TargetUniversalCRTVersion
property
will result in binaries that appear to be XP-compatible at first glance,
but exhibit erroneous behaviour due to a bug in the UCRT library used
by default (10.0.10240.0). Telltale symptoms will be POV-Ray resetting the
master povray.ini
back to its defaults, and complaining that it can't find
colors.inc
, each time it is started.
You really don't need to worry about any of the following section unless you want to do editor hacking ; the DLL's provided with the POV-Ray for Windows binaries are sufficient, and therefore you don't need to compile them at all if you don't want to.
The editor used in POV-Ray for Windows is based around a custom control
called CodeMax, supplied courtesy of Barry Allyn. Our implementation
consists of two DLL's: cmedit32.dll
and povcmax32.dll
(in this case,
as for the rest of the instructions, we will refer to the 32-bit
versions; if building for Win64 then just replace the '32' with '64').
If you are building debug versions, the DLL's will be suffixed with a
'd' (for example, cmedit32d.dll
).
The editor control itself is povcmax32.dll
. The source for povcmax is
not distributed with the AGPL3 version of POV-Ray. If you do wish to
obtain this source it may be found in prior releases (such as the
version 3.7 release candidates, which were not AGPL licensed). Please
note however that codemax is NOT AGPL and was only included with POV-Ray
under the terms of the former POV-Ray license.
The other DLL, cmedit32
, consists of support code that wraps and
enhances povcmax, and provides services such as file I/O, menus, state
storage/restore, and many other things. This DLL is built by the
'cmedit' project (which is by default disabled via configuration manager).
For reference, while most of the POV-Ray-specific customization is done
within cmedit, there are some POV-Ray specific modifications made to
codemax itself, particularly with respect to the code completion
support. However, for the most part, if you are intending to modify some
aspect of the POV-Ray interface to the editor, you will find you will
want to modify the contents of the cmedit project (or, possibly,
pvedit.cpp in the windows directory - this implements the API between
the POV-Ray executable and cmedit32.dll
). Note that if you alter this
API at all, you must change the editor version (EDITDLLVERSION
in
windows\pvedit.h
).
By default, POV-Ray will attempt to load the editor DLL's from the directory in which the EXE resides first. (The previous strategy was to look in the current working directory first; POV-Ray no longer checks the CWD, though you can force that behaviour if you wish by using the switch mentioned below).
If a DLL with the right name is found with the EXE it will be used;
otherwise, the default binary directory is used. If you are not sure
where it's looking, create the directory c:\temp
if it does not
already exist and then run pvengine
with the /debug
switch. Once it
has loaded have a look at c:\temp\povdebug.txt
, which will list the
search paths attempted.
There is a switch available on the command-line which can alter the
search path. If you specify /EDITDLLPATH
on the command-line and
follow it with a path, that path will be the only one checked. To force
POV-Ray to look in the current directory (i.e. the one the EXE is
launched from, which may not be the one the EXE is stored in), you can
pass .
as the edit DLL path.
Note: currently POV-Ray for Windows has no way of differentiating the
official editor DLL's from custom ones. We may at a future point provide
a means for unofficial DLL's to be loaded by the official PVENGINE
binary. In the meantime, if you want to make absolutely sure that your
unofficial binary doesn't load the official DLL's, and vice-versa, you
can rename the DLL's. Doing this requires you to tell the EXE of course -
to do this, open windows\pvengine.h
and look for where EDITDLLNAME
is defined. Just change that to suit, re-compile, and rename the DLL's
manually (or change the project to output different names).
If porting POV-Ray for Windows to another compiler, or even if you're just writing a makefile for the Visual Studio version, be aware that there is some special configuration in the project settings. The most critical one to replicate is the entry point - POVWIN uses a custom entry point, and unless you adjust the code, it is essential this is called prior to the C run-time library startup code. It's also highly likely that you will need to change the code that calls the RTL from within POV-Ray as well, unless you are using the Microsoft RTL (or a compatible one).
Currently, the entry point is POVWINStartup
, which is declared in
pvmem.cpp
. We use this to set up a private heap prior to any memory
allocation occurring (including even allocations done to construct any
global or static classes). If WIN_MEM_TRACKING
is defined (which it is
by default), pvmem.cpp
replaces the standard new
, new[]
, delete
,
and delete[]
operators with its own ones.
Apart from tracking the amount of memory used and freed, the above
tracking code, when _DEBUG
is defined, will also provide a custom
wrapper around each memory block which records where the new or malloc
occurred. By default, _DEBUG
also causes EXTRA_VALIDATION
to be
defined. When this is present, each block allocated has, in addition to
the standard extra fields added to the start, a single field of length
sizeof(ptrdiff_t)
added to the end of the block. This is then
initialized with a simple hash made up from various aspects of the
block, including its address, length, and the source line it was
allocated from. This is able to be checked at any time using the
Validate()
method to see if it has been altered since the block was
allocated (e.g. by an overrun).
Additionally, while it has not yet been implemented, the memory hooks give us the opportunity to provide access to a low-fragmentation heap or other useful memory allocation techniques (e.g. handling allocations differently during renders, if we want) for the purpose of improving performance.
The VFE library included with the source acts as a wrapper for the communications system ('POVMS') that connects the POV-Ray front and backend code (which are logically separate, and may even be on separate computers). POVMS is a message-passing system that is platform-neutral, thus allowing the backend to run on a different architecture than the frontend.
VFE provides a worker thread and various abstractions that handle most of the work of hooking a frontend up to the backend, even to the point where a frontend may be transient or stateless (i.e. HTTP, hence the term 'virtual'). The Windows and Unix versions of POV-Ray use VFE to connect the frontend to the backend.
If you wish to integrate POV-Ray into your own software (subject of course to the terms of the AGPL3), it is strongly recommended that you first study the source of povconsole.cpp, which implements a minimal example of how to setup and run a render. A more comprehensive example of console integration is in the unix support code, and for a GUI version, the windows support code should be read.
In either case, be sure to read through source/vfe/vfesession.h
;
it is heavily commented and is intended to be the
primary reference to the API.
If you have issues with getting VFE to work with your own custom frontend, please visit our support forums and ask about it: if there are any bugs or shortcomings we would like to fix them. We will also try to offer general assistance where possible, subject to time limitations.
If your project is commercial and you wish to have dedicated support, some of the POV-Ray developers may be available for contract work, depending on the circumstance. Please contact us for more information.
The POV-Team would like to thank all those who have assisted in bringing POV-Ray v3.8 and its predecessors to the public - you know who you are, and we salute you.