Commit Graph

2844 Commits

Author SHA1 Message Date
Dean Moldovan
bad1740213 Add checks to maintain a consistent Python code style and prevent bugs (#515)
A flake8 configuration is included in setup.cfg and the checks are
executed automatically on Travis:

* Ensures a consistent PEP8 code style
* Does basic linting to prevent possible bugs
2016-11-20 21:21:54 +01:00
Wenzel Jakob
df81546965 added forgotten initialization 2016-11-20 05:41:38 +01:00
Dean Moldovan
d079f41c26 Always use return_value_policy::move for rvalues (#510)
Fixes #509.

The move policy was already set for rvalues in PR #473, but this only
applied to directly cast user-defined types. The problem is that STL
containers cast values indirectly and the rvalue information is lost.
Therefore the move policy was not set correctly. This commit fixes it.

This also makes an additional adjustment to remove the `copy` policy
exception: rvalues now always use the `move` policy. This is also safe
for copy-only rvalues because the `move` policy has an internal fallback
to copying.
2016-11-20 05:31:02 +01:00
Wenzel Jakob
31fbf18ac7 replace redundant function_record->class_ field with 1 bit 2016-11-20 05:27:05 +01:00
Wenzel Jakob
7c2461eefd resolve issue involving inheritance + def_static + override (fixes #511) 2016-11-20 05:26:02 +01:00
Wenzel Jakob
405f6d1dfd make arithmetic operators of enum_ optional (#508)
Following commit 90d278, the object code generated by the python
bindings of nanogui (github.com/wjakob/nanogui) went up by a whopping
12%. It turns out that that project has quite a few enums where we don't
really care about arithmetic operators.

This commit thus partially reverts the effects of #503 by introducing
an additional attribute py::arithmetic() that must be specified if the
arithmetic operators are desired.
2016-11-17 23:24:47 +01:00
Lori A. Burns
99ddc9ac1a equals needed for test_copy_move_policies to build icpc 2016.3 (#507) 2016-11-17 11:01:11 +01:00
Dean Moldovan
4de271027d Improve consistency of array and array_t with regard to other pytypes
* `array_t(const object &)` now throws on error
* `array_t::ensure()` is intended for casters —- old constructor is
  deprecated
* `array` and `array_t` get default constructors (empty array)
* `array` gets a converting constructor
* `py::isinstance<array_T<T>>()` checks the type (but not flags)

There is only one special thing which must remain: `array_t` gets
its own `type_caster` specialization which uses `ensure` instead
of a simple check.
2016-11-17 08:55:42 +01:00
Dean Moldovan
c7ac16bb2e Add py::reinterpret_borrow<T>()/steal<T>() for low-level unchecked casts
The pytype converting constructors are convenient and safe for user
code, but for library internals the additional type checks and possible
conversions are sometimes not desired. `reinterpret_borrow<T>()` and
`reinterpret_steal<T>()` serve as the low-level unsafe counterparts
of `cast<T>()`.

This deprecates the `object(handle, bool)` constructor.

Renamed `borrowed` parameter to `is_borrowed` to avoid shadowing
warnings on MSVC.
2016-11-17 08:55:42 +01:00
Dean Moldovan
e18bc02fc9 Add default and converting constructors for all concrete Python types
* Deprecate the `py::object::str()` member function since `py::str(obj)`
  is now equivalent and preferred

* Make `py::repr()` a free function

* Make sure obj.cast<T>() works as expected when T is a Python type

`obj.cast<T>()` should be the same as `T(obj)`, i.e. it should convert
the given object to a different Python type. However, `obj.cast<T>()`
usually calls `type_caster::load()` which only checks the type without
doing any actual conversion. That causes a very unexpected `cast_error`.
This commit makes it so that `obj.cast<T>()` and `T(obj)` are the same
when T is a Python type.

* Simplify pytypes converting constructor implementation

It's not necessary to maintain a full set of converting constructors
and assignment operators + const& and &&. A single converting const&
constructor will work and there is no impact on binary size. On the
other hand, the conversion functions can be significantly simplified.
2016-11-17 08:55:42 +01:00
Dean Moldovan
b4498ef44d Add py::isinstance<T>(obj) for generalized Python type checking
Allows checking the Python types before creating an object instead of
after. For example:
```c++
auto l = list(ptr, true);
if (l.check())
   // ...
```
The above is replaced with:
```c++
if (isinstance<list>(ptr)) {
    auto l = reinterpret_borrow(ptr);
    // ...
}
```

This deprecates `py::object::check()`. `py::isinstance()` covers the
same use case, but it can also check for user-defined types:
```c++
class Pet { ... };
py::class_<Pet>(...);

m.def("is_pet", [](py::object obj) {
    return py::isinstance<Pet>(obj); // works as expected
});
```
2016-11-17 08:55:42 +01:00
Wenzel Jakob
281ccc692c exception constructor consistency improvements (fixes #492) 2016-11-16 17:59:56 +01:00
Sylvain Corlay
5027c4f95b Switch NumPy variadic indexing to per-value arguments (#500)
* Also added unsafe version without checks
2016-11-16 17:53:37 +01:00
Wenzel Jakob
2fb5f1d0c3 added message about recursion limits 2016-11-16 17:37:10 +01:00
Pim Schellart
90d27805b9 Extended enum support (#503)
* Allow enums to be ordered
* Support binary operators
2016-11-16 17:28:11 +01:00
Jason Rhinelander
2e76daa53f Enable testing for <optional> when available (#501) 2016-11-15 22:24:26 +01:00
Ivan Smirnov
425b4970b2 Add type casters for nullopt_t, fix none refcount (#499)
* Incref returned None in std::optional type caster

* Add type casters for nullopt_t

* Add a test for nullopt_t
2016-11-15 13:00:38 +01:00
Alexander Stukowski
9a110e6da8 Provide more control over automatic generation of docstrings (#486)
Added the docstring_options class, which gives global control over the generation of docstrings and function signatures.
2016-11-15 12:38:05 +01:00
Jason Rhinelander
617fbcfc1e Fix stl_bind to support movable, non-copyable value types (#490)
This commit includes the following changes:

* Don't provide make_copy_constructor for non-copyable container

make_copy_constructor currently fails for various stl containers (e.g.
std::vector, std::unordered_map, std::deque, etc.) when the container's
value type (e.g. the "T" or the std::pair<K,T> for a map) is
non-copyable.  This adds an override that, for types that look like
containers, also requires that the value_type be copyable.

* stl_bind.h: make bind_{vector,map} work for non-copy-constructible types

Most stl_bind modifiers require copying, so if the type isn't copy
constructible, we provide a read-only interface instead.

In practice, this means that if the type is non-copyable, it will be,
for all intents and purposes, read-only from the Python side (but
currently it simply fails to compile with such a container).

It is still possible for the caller to provide an interface manually
(by defining methods on the returned class_ object), but this isn't
something stl_bind can handle because the C++ code to construct values
is going to be highly dependent on the container value_type.

* stl_bind: copy only for arithmetic value types

For non-primitive types, we may well be copying some complex type, when
returning by reference is more appropriate.  This commit returns by
internal reference for all but basic arithmetic types.

* Return by reference whenever possible

Only if we definitely can't--i.e. std::vector<bool>--because v[i]
returns something that isn't a T& do we copy; for everything else, we
return by reference.

For the map case, we can always return by reference (at least for the
default stl map/unordered_map).
2016-11-15 12:30:38 +01:00
Wenzel Jakob
06bd27f536 import size_t into pybind11 namespace (fixes #498) 2016-11-15 06:37:39 +01:00
Jason Rhinelander
0780655808 Fix test compilation failure under gcc 4.9 (#496) 2016-11-13 10:41:31 +09:00
Jason Rhinelander
920e0e349d Add cmake option to override tests (#489)
When working on some particular feature, it's nice to be able to disable
all the tests except for the one I'm working on; this is currently
possible by editing tests/CMakeLists.txt, and commenting out the tests
you don't want.

This commit goes a step further by letting you give a list of tests you
do want when invoking cmake, e.g.:

    cmake -DPYBIND11_TEST_OVERRIDE="test_issues.cpp;test_pickling.cpp" ..

changes the build to build just those two tests (and changes the `pytest`
target to invoke just the two associated tests).

This persists in the build directory until you disable it again by
running cmake with `-DPYBIND11_TEST_OVERRIDE=`.  It also adds a message
after the pytest output to remind you that it is in effect:

    Note: not all tests run: -DPYBIND11_TEST_OVERRIDE is in effect
2016-11-13 09:10:53 +09:00
Wenzel Jakob
5e1c0445cf include backtrace in pybind11::detail::error_string (#494) 2016-11-12 16:57:30 +09:00
Jason Rhinelander
219b10ac99 Update travis and appveyor builds to eigen 3.3.0 (#491) 2016-11-11 16:15:38 +09:00
Wenzel Jakob
2b92a49115 updated check-style.sh docs 2016-11-08 10:58:22 +01:00
Wenzel Jakob
cc4efe69c2 more code style checks in Travis CI :) 2016-11-08 10:53:30 +01:00
Wenzel Jakob
fe40dfe67d address number caster regression (fixes #484) 2016-11-07 15:59:01 +01:00
Jason Rhinelander
c07ec31edf Don't construct unique_ptr around unowned pointers (#478)
If we need to initialize a holder around an unowned instance, and the
holder type is non-copyable (i.e. a unique_ptr), we currently construct
the holder type around the value pointer, but then never actually
destruct the holder: the holder destructor is called only for the
instance that actually has `inst->owned = true` set.

This seems no pointer, however, in creating such a holder around an
unowned instance: we never actually intend to use anything that the
unique_ptr gives us: and, in fact, do not want the unique_ptr (because
if it ever actually got destroyed, it would cause destruction of the
wrapped pointer, despite the fact that that wrapped pointer isn't
owned).

This commit changes the logic to only create a unique_ptr holder if we
actually own the instance, and to destruct via the constructed holder
whenever we have a constructed holder--which will now only be the case
for owned-unique-holder or shared-holder types.

Other changes include:

* Added test for non-movable holder constructor/destructor counts

The three alive assertions now pass, before #478 they fail with counts
of 2/2/1 respectively, because of the unique_ptr that we don't want and
don't destroy (because we don't *want* its destructor to run).

* Return cstats reference; fix ConstructStats doc

Small cleanup to the #478 test code, and fix to the ConstructStats
documentation (the static method definition should use `reference` not
`reference_internal`).

* Rename inst->constructed to inst->holder_constructed

This makes it clearer exactly what it's referring to.
2016-11-06 19:12:48 +01:00
Wenzel Jakob
e916d846bf minor: have enum::export_values() return a reference to *this as usual 2016-11-04 16:51:14 +01:00
Jason Rhinelander
f1b44a051a <optional> requires -std=c++17 (#479)
There are now more places than just descr.h that make use of these.
The new macro isn't quite the same: the old one only tested for a
couple features, while the new one checks for the __cplusplus version
(but doesn't even try to enable C++14 for MSVC/ICC).

g++ 7 adds <optional>, but including it in C++14 mode isn't allowed
(just as including <experimental/optional> isn't allowed in C++11 mode).
(This wasn't triggered in g++-6 because it doesn't provide <optional>
yet.)
2016-11-04 14:49:37 +01:00
Jason Rhinelander
dc0b4bd2c9 Add debugging info about .so size to build output (#477)
* Add debugging info about so size to build output

This adds a small python script to tools that captures before-and-after
.so sizes between builds and outputs this in the build output via a
string such as:

------ pybind11_tests.cpython-35m-x86_64-linux-gnu.so file size: 924696 (decrease of 73680 bytes = 7.38%)

------ pybind11_tests.cpython-35m-x86_64-linux-gnu.so file size: 998376 (increase of 73680 bytes = 7.97%)

------ pybind11_tests.cpython-35m-x86_64-linux-gnu.so file size: 998376 (no change)

Or, if there was no .so during the build, just the .so size by itself:

------ pybind11_tests.cpython-35m-x86_64-linux-gnu.so file size: 998376

This allows you to, for example, build, checkout a different branch,
rebuild, and easily see exactly the change in the pybind11_tests.so
size.

It also allows looking at the travis and appveyor build logs to get an
idea of .so/.dll sizes across different build systems.

* Minor libsize.py script changes

- Use RAII open
- Remove unused libsize=-1
- Report change as [+-]xyz bytes = [+-]a.bc%
2016-11-04 14:47:41 +01:00
Wenzel Jakob
45e6e6f6eb add note about custom type casters (fixes #480) 2016-11-04 11:06:22 +01:00
Jason Rhinelander
12edaaa66a Only enable std::optional if compiling in >= C++14 (#476) 2016-11-03 16:17:11 +01:00
Ivan Smirnov
44a69f78cf std::experimental::optional (#475)
* Add type caster for std::experimental::optional

* Add tests for std::experimental::optional

* Support both <optional> / <experimental/optional>

* Mention std{::experimental,}::optional in the docs
2016-11-03 13:42:46 +01:00
Wenzel Jakob
bd560acf40 smart pointer refcount fix by @dean0x7d with slight modifications (fixes #471) 2016-11-03 11:53:35 +01:00
Wenzel Jakob
0a9ef9c300 Merge pull request #472 from aldanor/feature/shared-dtypes
Support for sharing dtypes across extensions + public shared data API
2016-11-03 11:08:50 +01:00
Ivan Smirnov
cc8ff16547 Move register_dtype() outside of the template
(avoid code bloat if possible)
2016-11-03 09:35:05 +00:00
Ivan Smirnov
f95fda0eb2 Add docs re: shared data API 2016-11-03 09:35:05 +00:00
Ivan Smirnov
2dbf029705 Add public shared_data API
NumPy internals are stored under "_numpy_internals" key.
2016-11-03 09:35:05 +00:00
Ivan Smirnov
c546655dc2 Use pytest fixtures in numpy dtypes test module 2016-11-03 09:35:05 +00:00
Ivan Smirnov
2184f6d4d6 NumPy dtypes are now shared across extensions 2016-11-03 09:35:05 +00:00
Wenzel Jakob
a743ead455 Merge pull request #474 from aldanor/feature/numpy-dtype-ex
Overriding field names when binding structured dtypes
2016-11-03 09:44:30 +01:00
Ivan Smirnov
abd3429ce9 Add a test for numpy dtypes with custom names 2016-11-01 13:29:32 +00:00
Ivan Smirnov
e8b50360fe Add dtype binding macro that allows setting names
PYBIND11_NUMPY_DTYPE_EX(Type, F1, "N1", F2, "N2", ...)
2016-11-01 13:27:35 +00:00
Dean Moldovan
03f627ebb1 Make reference(_internal) the default return value policy for properties (#473)
* Make reference(_internal) the default return value policy for properties

Before this, all `def_property*` functions used `automatic` as their
default return value policy. This commit makes it so that:

 * Non-static properties use `reference_interal` by default, thus
   matching `def_readonly` and `def_readwrite`.

 * Static properties use `reference` by default, thus matching
   `def_readonly_static` and `def_readwrite_static`.

In case `cpp_function` is passed to any `def_property*`, its policy will
be used instead of any defaults. User-defined arguments in `extras`
still have top priority and will override both the default policies and
the ones from `cpp_function`.

Resolves #436.

* Almost always use return_value_policy::move for rvalues

For functions which return rvalues or rvalue references, the only viable
return value policies are `copy` and `move`. `reference(_internal)` and
`take_ownership` would take the address of a temporary which is always
an error.

This commit prevents possible user errors by overriding the bad rvalue
policies with `move`. Besides `move`, only `copy` is allowed, and only
if it's explicitly selected by the user.

This is also a necessary safety feature to support the new default
return value policies for properties: `reference(_internal)`.
2016-11-01 11:44:57 +01:00
Wenzel Jakob
030d10e826 minor style fix 2016-10-28 01:23:42 +02:00
Wenzel Jakob
496feacfd0 pybind11: implicitly convert NumPy integer scalars
The current integer caster was unnecessarily strict and rejected
various kinds of NumPy integer types when calling C++ functions
expecting normal integers. This relaxes the current behavior.
2016-10-28 01:02:46 +02:00
Wenzel Jakob
a6f85eb985 Merge pull request #465 from jagerman/prevent-object-overwriting
Prevent overwriting previous declarations
2016-10-25 10:03:53 +02:00
Jason Rhinelander
6873c202b3 Prevent overwriting previous declarations
Currently pybind11 doesn't check when you define a new object (e.g. a
class, function, or exception) that overwrites an existing one.  If the
thing being overwritten is a class, this leads to a segfault (because
pybind still thinks the type is defined, even though Python no longer
has the type).  In other cases this is harmless (e.g. replacing a
function with an exception), but even in that case it's most likely a
bug.

This code doesn't prevent you from actively doing something harmful,
like deliberately overwriting a previous definition, but detects
overwriting with a run-time error if it occurs in the standard
class/function/exception/def registration interfaces.

All of the additions are in non-template code; the result is actually a
tiny decrease in .so size compared to master without the new test code
(977304 to 977272 bytes), and about 4K higher with the new tests.
2016-10-24 22:45:51 -04:00
Wenzel Jakob
dd9bd7778f Merge pull request #453 from aldanor/feature/numpy-scalars
NumPy scalars to ctypes conversion support
2016-10-25 01:15:25 +02:00