Code contributions to OSL are always welcome. That's a big part of why it's an open source project. Please review this document to get a briefing on our process.
Contributors should be reading the osl-dev mail list:
You can sign up for the mail list on your own using the link above.
We also have an openshadinglanguage channel on the
ASWF Slack. But note that the Slack channel is
not archived, so while it's ok for fleeting comments, it is unreliable as
a permanent repository of useful information. Serious discussions that are
we would want to refer to over time should be sent to the mail list instead.
We use GitHub's issue tracking system for reporting bugs and requesting enhancements: https://github.com/AcademySoftwareFoundation/OpenShadingLanguage/issues
If you are merely asking a question ("how do I..."), please do not file an issue, but instead ask the question on the OSL developer mail list.
If you are submitting a bug report, please be sure to note which version of OSL you are using, on what platform (OS/version, which compiler you used, and any special build flags or other unusual environmental issues). Please give an account of
- what you tried
- what happened
- what you expected to happen instead
with enough detail that others can reproduce the problem.
To protect the project -- and the contributors! -- we do require a Contributor License Agreement (CLA) for anybody submitting changes.
-
Corporate CLA If you are writing the code as part of your job, or if there is any possibility that your employers might think they own any intellectual property you create. This needs to be executed by someone who has signatory power for the company.
-
Individual. If you are an individual writing the code on your own time and you're SURE you are the sole owner of any intellectual property you contribute.
The easiest way to sign CLAs is digitally using EasyCLA. Companies who prefer not to use the online tools may sign, scan, and email the executed copy to manager@lfprojects.org.
The CLA allows a company to name a "CLA Manager" (who does not need signatory power) who has the ability to use the online system to add or delete individual employees of the company who are authorized to submit pull requests, without needing to get an executive to amend and sign the agreement each time.
Please note that these CLAs are based on the Apache 2.0 CLAs, and differ minimally, only as much as was required to correctly describe the EasyCLA process and our use of a CLA manager.
** Contribution sign off
ASWF requires the use of the Developer’s Certificate of Origin 1.1 (DCO), which is the same mechanism that the Linux® Kernel and many other communities use to manage code contributions. The DCO is considered one of the simplest tools for sign offs from contributors as the representations are meant to be easy to read and indicating signoff is done as a part of the commit message.
Here is an example Signed-off-by line, which indicates that the submitter accepts the DCO:
Signed-off-by: John Doe <john.doe@example.com>
You can include this automatically when you commit a change to your local
git repository using git commit -s. You might also want to
leverage this command line tool for
automatically adding the signoff message on commits.
The best way to submit changes is via GitHub Pull Request. GitHub has a Pull Request Howto.
All code must be formally reviewed before being merged into the official repository. The protocol is like this:
-
Get a GitHub account, fork AcademySoftwareFoundation/OpenShadingLanguage to create your own repository on GitHub, and then clone it to get a repository on your local machine.
-
Edit, compile, and test your changes.
-
Push your changes to your fork (each unrelated pull request to a separate "topic branch", please).
-
Make a "pull request" on GitHub for your patch.
-
If your patch will induce a major compatibility break, or has a design component that deserves extended discussion or debate among the wider OSL community, then it may be prudent to email osl-dev pointing everybody to the pull request URL and discussing any issues you think are important.
-
The reviewer will look over the code and critique on the "comments" area, or discuss in email. Reviewers may ask for changes, explain problems they found, congratulate the author on a clever solution, etc. But until somebody says "LGTM" (looks good to me), the code should not be committed. Sometimes this takes a few rounds of give and take. Please don't take it hard if your first try is not accepted. It happens to all of us.
-
After approval, one of the senior developers (with commit approval to the official main repository) will merge your fixes into the master branch.
There are two overarching rules:
-
When making changes, conform to the style and conventions of the surrounding code.
-
Strive for clarity, even if that means occasionally breaking the guidelines. Use your head and ask for advice if your common sense seems to disagree with the conventions.
Below we try to enumerate the guidelines embodied in the code.
C++ implementation should be named *.cpp. Headers should be named .h.
All headers should contain
#pragma once
All source files should begin with the copyright and license, which can be found in the LICENSE file (or cut and pasted from any other other source file). Two notes on this:
-
For NEW source files, please change the copyright year to the present. DO NOT edit existing files only to update the copyright year, it just creates pointless deltas and offers no increased protection.
-
Occasionally a file contains substantial code from another project and will also list its copyright. Do NOT copy that notice to any new files, it really only applies to the particular file in which it appears.
We use clang-format to enforce formatting rules for the vast majority of the code base. It is automatically run during CI and is a CI failure if the code does not conform. There are, however, parts of the code base that are not clang-formatted, either because we have not yet put them under the clang-format regime, or because individual modules were deemed to be less clear with clang-format than with hand formatting.
For non-clang-format regions of code, NEVER alter somebody else's code to reformat just because you found something that violates the rules. Let the group/author/leader know, and resist the temptation to change it yourself.
Each line of text in your code should be at most 80 characters long. Exceptions are allowed for those rare cases where letting a line be longer (and wrapping on an 80-character window) is actually a better and clearer alternative than trying to split it into two lines. Sometimes this happens, but it's extremely rare.
Indent 4 spaces at a time, and use actual spaces, not tabs. For files that must use tabs for some reason, tabs should always be on 8's. Most editors have a setting that forces indentations to be spaces. With emacs, you can do this:
(setq c-default-style "bsd")
(setq-default indent-tabs-mode nil)
Opening brace on the same line as the condition or loop.
One statement per line, except in rare cases where violating this rule makes the code more clear.
Three (3) consecutive empty lines between function or class method implementations, one blank line between method declarations within a class definition. Put a single blank line within a function if it helps to visually separate different sequential tasks. Don't put multiple blank lines in a row within a function, or blank lines right after an opening brace or right before a closing brace. The goal is to use just enough white space to help developers visually parse the code (for example, spotting at a glance where new functions begin), but not so much as to spread it out or be confusing.
For if, for, while, etc., put a space before the paren, but NOT inside the parens. For example:
if (foo) // Yes
if(foo) // No
if ( foo ) // No
Function calls should have a space between the function name and the opening parenthesis, NO space inside the parenthesis, except for a single blank space between each argument. For example:
x = foo(a, b); // Yes
x = foo ( a, b ); // No
x = foo (a,b); // No
Function declarations: function names should begin at column 0 for a full function definition. (It's ok to violate this rule for very short inline functions within class definitions.)
Here is a short code fragment that shows some of these rules in action:
static int
function(int a, int b)
{
int x = a + b;
if (a == 0 || b == 0) {
x += 1;
x *= 4;
} else {
x -= 3;
}
for (int i = 0; i < 3; ++i) {
x += a * i;
x *= foo(i); // function call
}
return x;
}
Don't ever reformat, re-indent, change whitespace, or make any other such changes to working code. If you're adding just a few lines or fixing a bug in existing code, stick to the style of the surrounding code. In very rare circumstances, and with consensus of the group, reformatting is sometimes helpful to improve code readability, but it should be done as a separate formatting-only checkin.
In general, classes and templates should start with upper case and capitalize new words:
class CustomerList;
In general, local variables should start with lower case.
If your class is extremely similar to, or modeled after, something in the standard library, Boost, or something else we interoperate with, it's ok to use their naming conventions. For example, very general utility classes and templates (the kind of thing you would normally find in std or boost) should be lower case with underscores separating words, as they would be if they were standards.
template <class T> shared_ptr;
class scoped_mutex;
Macros should be ALL_CAPS, if used at all.
Names of data should generally be nouns. Functions/methods are trickier: a the name of a function that returns a value but has no side effects should be a noun, but a procedure that performs an action should be a verb.
Try to avoid public data members, although there are some classes that serve a role similar to a simple C struct -- a very straightforward collection of data members. In these, it's fine to make the data members public and have clients set and get them directly.
Private member data should be named m_foo (alternately, it's ok to use the common practice of member data foo_, but don't mix the conventions within a class).
Private member data that needs public accessors should use the convention:
void foo (const T& newfoo) { m_foo = newfoo; }
const T& foo () const { return m_foo; }
Avoid multiple inheritance.
Namespaces: yes, use them!
Prefer C++11 std rather than Boost or other third party libraries, where
both can do the same task.
If you see a third party library already used as a dependency (such as OIIO, Boost, Ilmbase, or LLVM), feel free to any of its public features in OSL internals (provided those features are present in the minimum version of that library that we support).
Please do not add any NEW dependencies without debate on osl-dev and approval of the project leader.
Use these libraries for OSL internals, but please DO NOT let them infect any of our public APIs unless it's been thoroughly discussed and approved by the group. (Exceptions: it's ok to use OIIO and Imath classes in our public APIs.)
Comment philosophy: try to be clear, try to help teach the reader what's going on in your code.
Prefer C++ comments (starting line with //) rather than C comments (/* ... */).
For any function that may be used by other programmers (e.g., public or protected members of classes), please use Doxygen-style comments. They looks like this:
/// Explanation of a class. Note THREE slashes!
/// Also, you need at least two lines like this. If you don't have enough
/// for two lines, make one line blank like this:
///
class myclass {
....
float foo; ///< Doxygen comments on same line look like this
}
If you know Doxygen well, please feel free to use the various other markups. But don't go so crazy with Doxygen markups that the comment itself, in an ordinary editor, is not as readable as possible. The Doxygen-generated pages are nice, but the place that needs to be most readable is in the code.
Macros should be used only rarely -- prefer inline functions, templates, enums, or "const" values wherever possible.
When in doubt, look elsewhere in the code base for examples of similar structures and try to format your code in the same manner.