summary |
shortlog | log |
commit |
commitdiff |
tree
first ⋅ prev ⋅ next
Robert Bradshaw [Tue, 25 Sep 2007 21:47:12 +0000 (14:47 -0700)]
Use assignable_from for conditional node
Fixes
cdef int a, b
cdef int *foo = &a if a else &b
Robert Bradshaw [Wed, 19 Sep 2007 22:35:13 +0000 (15:35 -0700)]
Use PY_VERSION_HEX rather than sys.version_info for typeslot versioning.
so the generated c files don't depend on the version of python used to run Cython
Thomas Hunger [Wed, 19 Sep 2007 21:15:34 +0000 (23:15 +0200)]
Add versioning to slots so that additional
slots for new pyton versions are not emitted
for older versions. This stops compiler warnings.
Robert Bradshaw [Wed, 19 Sep 2007 20:26:34 +0000 (13:26 -0700)]
Fix classmethod for use in cdef classes.
The issue here (as pointed out by Thomas Hunger) is that once the CClass
body gets executed the class is already constructed, and its methods
are actual methods not functions (as in Python).
The CClassScope now returns a utility function on lookup of "classmethod"
that creates a class method out of a method (rather than out of a function).
When added to the type dictionary, the result is exactly the same as setting
the METH_CLASS flag.
Thomas Hunger [Wed, 19 Sep 2007 13:40:17 +0000 (15:40 +0200)]
Changed the assigment code generator to emit
a direct change of a classes tp_dict instead
of SetAttr if assignment code is executed in
a class body.
To tell the code generator that we have a class
entry, entry has a new attribute "is_member".
Thomas Hunger [Wed, 19 Sep 2007 10:47:06 +0000 (12:47 +0200)]
Emit and release temporary variables in global namespace
because the class-body code is exectuted at module-init time.
Robert Bradshaw [Wed, 19 Sep 2007 01:56:46 +0000 (18:56 -0700)]
Finish class members
Robert Bradshaw [Wed, 19 Sep 2007 00:56:11 +0000 (17:56 -0700)]
Class members in cdef classes
tom@tom [Tue, 18 Sep 2007 17:43:44 +0000 (19:43 +0200)]
partial code for execution in classes
Stefan Behnel [Tue, 11 Sep 2007 19:12:49 +0000 (21:12 +0200)]
use 'with GIL' instead of 'withGIL' to avoid introducing a non-Python keyword and to support later extension for other contexts
Stefan Behnel [Mon, 10 Sep 2007 21:00:48 +0000 (23:00 +0200)]
support automatic GIL grabbing inside a function by addign 'withGIL' to its signature
Robert Bradshaw [Tue, 11 Sep 2007 23:39:51 +0000 (16:39 -0700)]
Type check after (not before) operation on inplace assignment.
Robert Bradshaw [Mon, 10 Sep 2007 17:35:20 +0000 (10:35 -0700)]
enable attribute access of cdef functions
Robert Bradshaw [Fri, 7 Sep 2007 09:49:43 +0000 (02:49 -0700)]
Check return values before PyErr_Occurred() for coercion nodes.
Robert Bradshaw [Wed, 5 Sep 2007 23:43:04 +0000 (16:43 -0700)]
incref_local_binop option for SAGE
This is so mutating inplace operations can be detected safely.
Robert Bradshaw [Wed, 5 Sep 2007 18:39:47 +0000 (11:39 -0700)]
Better error for ** on integer types.
Stefan Behnel [Mon, 3 Sep 2007 14:17:12 +0000 (16:17 +0200)]
do not use __builtin_expect with GCC < 2.96
Stefan Behnel [Mon, 3 Sep 2007 14:13:26 +0000 (16:13 +0200)]
rewrote Python 2.5-ism
Stefan Behnel [Mon, 3 Sep 2007 18:07:01 +0000 (20:07 +0200)]
fixed typo
Stefan Behnel [Sat, 25 Aug 2007 17:01:14 +0000 (19:01 +0200)]
fix for property functions
Stefan Behnel [Mon, 30 Jul 2007 20:18:23 +0000 (22:18 +0200)]
extract FQ module name from distutils
Robert Bradshaw [Wed, 29 Aug 2007 21:54:29 +0000 (14:54 -0700)]
version number 0.9.6.5
Robert Bradshaw [Sun, 19 Aug 2007 11:14:41 +0000 (04:14 -0700)]
avoid argument parsing (via meth_o, meth_noargs) for non-python-object arguments, upgrade version number to 0.9.6.4
compiles and runs SAGE fine
Robert Bradshaw [Sun, 19 Aug 2007 10:56:31 +0000 (03:56 -0700)]
remove pointer warning for generic arguments
Robert Bradshaw [Sun, 19 Aug 2007 07:44:50 +0000 (00:44 -0700)]
meth_o and meth_noargs for module-level functions
previous patch was just for classes
Robert Bradshaw [Sun, 19 Aug 2007 02:44:28 +0000 (19:44 -0700)]
Avoid call to PyArg_ParseTupleAndKeywords for zero- and single-argument functions.
Robert Bradshaw [Sun, 19 Aug 2007 00:40:39 +0000 (17:40 -0700)]
PyObject -> Py_ssize_t now uses __index__ rather than __int__ (even in function signatures)
Robert Bradshaw [Fri, 17 Aug 2007 23:11:06 +0000 (16:11 -0700)]
Correct flags on special methods so one can call their python functions directly.
Robert Bradshaw [Fri, 17 Aug 2007 22:46:16 +0000 (15:46 -0700)]
Docstrings for special methods
Robert Bradshaw [Fri, 17 Aug 2007 22:06:54 +0000 (15:06 -0700)]
version number
Robert Bradshaw [Fri, 17 Aug 2007 08:47:56 +0000 (01:47 -0700)]
Raise error when declaring reserved names.
William Stein [Sun, 29 Jul 2007 16:38:20 +0000 (09:38 -0700)]
Fixed some typos reported by Helmut Jarausch
Robert Bradshaw [Sun, 29 Jul 2007 01:30:36 +0000 (18:30 -0700)]
remove gcc dependance for inline, branch prediction
William Stein [Sat, 28 Jul 2007 20:48:10 +0000 (13:48 -0700)]
More Pyrex --> Cython conversions.
William Stein [Sat, 28 Jul 2007 20:42:16 +0000 (13:42 -0700)]
A few more Pyrex --> Cython changes.
William Stein [Sat, 28 Jul 2007 20:40:10 +0000 (13:40 -0700)]
Changing Python. to Cython.
William Stein [Sat, 28 Jul 2007 20:39:28 +0000 (13:39 -0700)]
Changing from pyrex/sagex to cython -- snapshot.
William Stein [Sat, 28 Jul 2007 20:18:27 +0000 (13:18 -0700)]
Merge in lxml-pyrex enum patch.
William Stein [Sat, 28 Jul 2007 20:16:04 +0000 (13:16 -0700)]
Merge in C/API patch from lxml-pyrex.
Robert Bradshaw [Wed, 11 Jul 2007 05:02:54 +0000 (22:02 -0700)]
Added pre-import option, which assumes all undeclared names come from the given module.
Emulates behavior of
from <module> import *
Robert Bradshaw [Tue, 10 Jul 2007 22:42:24 +0000 (15:42 -0700)]
gcc branch prediction
For some reason, it actually makes it slower in some of the most common cases (now commented out). Why? It does help in others.
Robert Bradshaw [Tue, 10 Jul 2007 09:55:12 +0000 (02:55 -0700)]
fix incompatable tmp release between 0.9.5.1a and fast tuple unpacking
Robert Bradshaw [Tue, 10 Jul 2007 09:15:10 +0000 (02:15 -0700)]
latest 0.9.4 changes into 0.9.5 branch
Robert Bradshaw [Tue, 10 Jul 2007 09:08:20 +0000 (02:08 -0700)]
finish merging, fix weakref segfault, all SAGE doctests pass
Robert Bradshaw [Sat, 7 Jul 2007 09:01:00 +0000 (02:01 -0700)]
Jim Kleckner patch to allow trailing commas in function argument lists
Robert Bradshaw [Sat, 7 Jul 2007 08:56:57 +0000 (01:56 -0700)]
sagex_multi_file.patch from Joel Mohler (sane cross-file type cmp)
Robert Bradshaw [Sat, 7 Jul 2007 07:58:33 +0000 (00:58 -0700)]
Fast tuple unpacking
Robert Bradshaw [Thu, 7 Jun 2007 04:26:36 +0000 (21:26 -0700)]
fix inplace operators for 0.9.5.1a
Robert Bradshaw [Thu, 7 Jun 2007 04:16:25 +0000 (21:16 -0700)]
copy ModuleNode changes to new file
Robert Bradshaw [Thu, 7 Jun 2007 04:01:58 +0000 (21:01 -0700)]
initial merge
Robert Bradshaw [Thu, 7 Jun 2007 02:22:57 +0000 (19:22 -0700)]
Pyrex Official version 0.9.5.1a
Robert Bradshaw [Thu, 7 Jun 2007 02:02:58 +0000 (19:02 -0700)]
type narrowing for pxd files
Robert Bradshaw [Fri, 1 Jun 2007 00:47:16 +0000 (17:47 -0700)]
type narrowing for cdef methods
Now if you inherit cdef methods from another class, you may re-declare the
arguments and return variables to be sub-types of the original declared type.
This will be especially convenient for the SAGE arithmetic architecture.
Type-checking is performed if necessary.
Robert Bradshaw [Wed, 9 May 2007 07:31:34 +0000 (00:31 -0700)]
fix semantics of cmp() -- should not be a bint
Robert Bradshaw [Fri, 4 May 2007 20:05:53 +0000 (13:05 -0700)]
Added warning about infinite loops in for-from-by statement.
One rarely, if ever, wants the [step] argument to be negative.
Robert Bradshaw [Fri, 4 May 2007 19:44:30 +0000 (12:44 -0700)]
Compile-time checking for builtin names, cast TypeObjects to PyObject* so c compiler won't complain.
Robert Bradshaw [Thu, 26 Apr 2007 05:04:07 +0000 (22:04 -0700)]
comments, builtin objects
Robert Bradshaw [Thu, 26 Apr 2007 03:42:45 +0000 (20:42 -0700)]
Added for i from ... ['by' step] syntax.
E.g.
for i from 0 <= i < 10 by 2:
print i
0
2
4
6
8
Old for-from loops remain exactly the same (using the ++ or --). If step is specified, the increment operator will be += step or -= step, depending on the orientation of the inequalities (as before).
NOTE: 'by' is now a keyword
Robert Bradshaw [Thu, 26 Apr 2007 02:54:15 +0000 (19:54 -0700)]
Added a bint c type, which is a c int that coerces to and from python objects via the boolean routines.
The purpose of this type is to free the coder from having to use
bool() when retrieving and returning semantically "boolean" values
(e.g. the result of a compare).
The bint type is a subclass of the int type, and the only difference
is that it uses PyBool_FromLong and PyObject_IsTrue rather than
PyInt_FromLong and PyInt_AsLong. Arithmatic on bints will return ints.
Where it makes sense, several builtin functions have been re-declared
to return bints, as well as comparisons and the boolean operations
or, and, and not.
Robert Bradshaw [Sat, 14 Apr 2007 11:13:05 +0000 (04:13 -0700)]
Cache __builtin__ name lookups so they are performed on module load rather than at every use.
The code "__Pyx_GetName(__pyx_b, __pyx_n_[string])" is performed in several
thousand places throughout the sage library, and can be quite expensive
(a dictionary lookup, possibly raising an error, etc.) This is redundant
as the result will always be the same. I perform the lookup once
(on loading the module), then have a pointer to the result for all subsequent use.
The most common examples are bool/str/int (both as function calls and
in isinstance), True/False, and raisign errors.
A side feature is that on loading a module with an illegal __builtin__ name,
it will complain at load time rather than at run time.
Robert Bradshaw [Tue, 27 Feb 2007 11:46:35 +0000 (03:46 -0800)]
Inline keyword for cdef functions, variable assignment on declaration (+optimization)
"cdef inline foo()" now valid, and will place inline in the resulting c code
"cdef o = expr" and "cdef type x = expr" now valid.
This may not seem like a huge change, but it ended up requiring
quite a bit of work. The variables are still all declared at the
top, but the assignment takes place at the specified line in the
code.
If an assignment is made at declaration, the variable is initalized to
0 rather than None (also skipping an INCREF) and Py_XDECREF is used on
exiting the function (in case an error occured before the actual value
was calculated). Hence these variables MUST NOT be used before they are
defined or it will probably segfault.
Robert Bradshaw [Sat, 24 Feb 2007 12:44:11 +0000 (04:44 -0800)]
disable builtin functions that conflict with type names
add dummy py_index for cdef arrays so subexprs valid
Robert Bradshaw [Sat, 24 Feb 2007 11:23:45 +0000 (03:23 -0800)]
Fix sizeof for dotted (cimported) types
Robert Bradshaw [Fri, 23 Feb 2007 21:46:02 +0000 (13:46 -0800)]
modify sizeof() operator to return size of extension type struct
Robert Bradshaw [Fri, 23 Feb 2007 08:18:05 +0000 (00:18 -0800)]
Figured out how to use the Python/C API for some builtin functions (such as len) to avoid python calling conventions.
Robert Bradshaw [Fri, 23 Feb 2007 04:52:31 +0000 (20:52 -0800)]
Conditional expressions
Changes in grammar required change for this, see http://www.python.org/dev/peps/pep-0308/
Most noteably for list comprehensions (dissambiguate the if)
Robert Bradshaw [Thu, 22 Feb 2007 21:17:25 +0000 (13:17 -0800)]
fast cdef type indexing into lists/tuples via runtime type checking
- 5 times as fast for L[i], list/tuple L, cdef integer type i, 0 <= i < len(L)
- no change for L[a], python object a
- 1-1.5% slowdown for L[i], i negative or out of range
Robert Bradshaw [Wed, 21 Feb 2007 06:17:26 +0000 (22:17 -0800)]
Disable a**b for cdef int's. (Return result was a double.)
Robert Bradshaw [Sat, 27 Jan 2007 07:06:06 +0000 (23:06 -0800)]
De-allocate function temp variables _after_ computing return value, in case an exception is thrown, caught, and said temp variables still need to be accessed.
Robert Bradshaw [Wed, 17 Jan 2007 04:24:21 +0000 (20:24 -0800)]
fix temp allocation order, remove straggling comment output
Robert Bradshaw [Tue, 16 Jan 2007 02:21:26 +0000 (18:21 -0800)]
FloorDiv operation
Robert Bradshaw [Tue, 16 Jan 2007 01:39:47 +0000 (17:39 -0800)]
Implemented inplace arithmetic
William Stein [Thu, 11 Jan 2007 06:25:21 +0000 (22:25 -0800)]
Get rid of this sort of error: "Cannot assign type 'gsl_complex' to 'gsl_complex'"
The solution in this patch is somewhat hackish, but should be OK.
Robert Bradshaw [Wed, 10 Jan 2007 09:06:30 +0000 (01:06 -0800)]
List comprehension
William Stein [Mon, 18 Dec 2006 02:59:29 +0000 (18:59 -0800)]
Peter Johnson (peter@tortall.net) weakref patch
I recently ran into this problem myself (as the current code causes
Python to crash), so I whipped up a quick patch that fixes it for me.
I think it follows all of the weakref guidelines now.
The patch is against the LXML svn pyrex
(http://codespeak.net/svn/lxml/pyrex/). Is there a different SVN I
should be pointing to? It patches functions generate_new_function,
generate_dealloc_function, generate_traverse_function, and
generate_clear_function.
The patch just compares the entry.name against "__weakref__"; this
could probably be centralized in the Entry object if so desired.
William Stein [Mon, 18 Dec 2006 02:05:12 +0000 (18:05 -0800)]
Apply Nick Alexander's patch so that Sagex that embeds positions in module, cdef class, and def class docstrings.
William Stein [Thu, 30 Nov 2006 19:57:57 +0000 (11:57 -0800)]
Eric Huss's readonly variable patch:
There's a minor bug, if you try to cdef a readonly variable at the module
scope, the compiler raises an UnboundLocalError exception. The attached
patch seems to clear up the problem and gets the compiler to report the
correct error.
William Stein [Sat, 25 Nov 2006 09:26:53 +0000 (01:26 -0800)]
Modified usage banner.
William Stein [Fri, 3 Nov 2006 23:41:38 +0000 (15:41 -0800)]
Had to also put the type code in for the signed case.
William Stein [Fri, 3 Nov 2006 23:34:50 +0000 (15:34 -0800)]
Made some error output more verbose.
William Stein [Fri, 3 Nov 2006 23:34:39 +0000 (15:34 -0800)]
Added support for automatic conversion of input types to Py_ssize_t.
William Stein [Fri, 3 Nov 2006 17:33:27 +0000 (09:33 -0800)]
Made annoted C files easier to read.
William Stein [Fri, 3 Nov 2006 04:29:29 +0000 (20:29 -0800)]
Fix one warning was too high a level
William Stein [Thu, 2 Nov 2006 15:46:17 +0000 (07:46 -0800)]
Made the output generated by Pyrex much more verbose with better context information.
For example:
/* "/Volumes/HOME/s/devel/sage-1/sage/matrix/matrix_generic_sparse.pyx":581
x = set(v.keys()).intersection(set(w.keys()))
a = 0
for k in x: # <<<<<<<<<<<<<<
a = a + v[k]*w[k]
return a
*/
__pyx_3 = PyObject_GetIter(__pyx_v_x); if (!__pyx_3) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 581; goto __pyx_L1;}
for (;;) {
William Stein [Sun, 29 Oct 2006 01:21:12 +0000 (18:21 -0700)]
Changed that error to a warning.
William Stein [Sun, 29 Oct 2006 01:19:51 +0000 (18:19 -0700)]
Fixed an error in a call to the error function.
William Stein [Fri, 27 Oct 2006 11:22:40 +0000 (06:22 -0500)]
Fix a Py_ssize_t build problem.
William Stein [Wed, 25 Oct 2006 07:05:47 +0000 (02:05 -0500)]
Add a new option "-e" or "--embed-positions" to Pyrex.
-p, --embed-positions If specified, the positions in Pyrex files of each
function definition is embedded in its docstring.
This is very useful to support interactive viewing of *Pyrex* source
code in, e.g, IPython.
William Stein [Wed, 25 Oct 2006 06:21:53 +0000 (01:21 -0500)]
Change so that warning for multiple declarations are printed by default.
William Stein [Mon, 23 Oct 2006 03:51:53 +0000 (22:51 -0500)]
Delete spurious print statement (that was used for debugging).
William Stein [Sun, 22 Oct 2006 04:56:20 +0000 (21:56 -0700)]
Add correct setting of tp_name to the full module name in Nodes.py.
This required a number of changes to a few files. Basically, the
full module name is determined in Main.py. It is then passed around
a bit until it is used when generating tp_name.
This change was needed because otherwise pickling of extension classes
with full module names like sage.rings.integer.Integer would fail
(since Python would look for integer.Integer instead). NOTE: This is
pickling of the extension class itself, not of instances (which could
also fail, because the class doesn't pickle).
William Stein [Sun, 22 Oct 2006 04:54:06 +0000 (21:54 -0700)]
Print out the usage message if pyrexc is called with no options and no input source files.
William Stein [Sun, 22 Oct 2006 03:56:56 +0000 (20:56 -0700)]
Don't list experimental macosx only options. Never needed in context of SAGE.
William Stein [Sun, 22 Oct 2006 03:55:47 +0000 (20:55 -0700)]
Add to help that multiple include directories are allowed.
William Stein [Sun, 22 Oct 2006 02:47:08 +0000 (19:47 -0700)]
Added nice error messages with context information.
Finally, you can see the line itself where the error occured,
instead of just the line number!!
William Stein [Sun, 22 Oct 2006 02:35:04 +0000 (19:35 -0700)]
Print "warning: " before warnings, so they look different than errors.
William Stein [Sun, 22 Oct 2006 01:53:34 +0000 (18:53 -0700)]
Added a "Warning" class, and changed it so redeclaring or re-importing is a warning rather than an error.
Because Pyrex has no #ifndef macro, it is impossibly painful to use
pxi files for declarations in a large project. SAGE is a large project.
Also, in Python it is not an error to import a module twice. Thus
more in line with Python's behavior, multiple declarations of the same
symbol is no longer an error.
William Stein [Fri, 20 Oct 2006 04:30:50 +0000 (21:30 -0700)]
Version of Pyrex distributed with SAGE on October 19, 2006
William Stein [Fri, 20 Oct 2006 04:28:18 +0000 (21:28 -0700)]
Official Pyrex version 0.9.4.1