|
Fixed memory allocation for list fragments. |
|
|
Merged branches, added a separate test of xrange and lists. |
|
|
Fixed scope handling where names are imported into namespaces. |
|
|
Merged dynamic/nested function and general development branches. |
|
|
Replaced the native mechanism for list initialisation with a Python-based |
|
|
Introduced the notion of dynamic definitions, to be used in conjunction with the |
|
|
Changed the function/lambda support to generate special function instances for |
|
|
Changed parameter initialisation to explicitly use Instance() instead of None. |
|
|
Fixed function name to avoid conflict with keyword. |
|
|
Added a test of nested functions. |
|
|
Added null argument notation in order to actually support other arguments when |
|
|
Removed usage of getattr by introducing lambda functions providing the methods |
|
|
Added a missing operator definition to micropython.common.augassign_methods. |
|
|
Added node qualification of name usage, since some name usage really involves |
|
|
Moved common functionality into a superclass, introducing generic raw image |
|
|
Tidied the output further, adding CSS class overrides for certain types of name. |
|
|
Added the _def attribute to AST nodes in order to refer to inspected objects. |
|
|
Changed the initialisation of Summariser instances. |
|
|
Moved a test into the failure category. |
|
|
Fixed program initialisation. |
|
|
Fixed errors when overspecific attribute usage caused the removal of attributes, |
|
|
Removed links to "" in the graph output. |
|
|
Attempted to make the API more regular. |
|
|
Added an elementary mechanism for estimating the cost of executed instructions, |
|
|
Prevented propagation of mutable users information. |
|
|
Introduced loop nodes as attribute users tracking attribute usage. |
|
|
Fixed while loop test boolean evaluation. |
|
|
Simplified the usage tracking by merely recording active users (providers) of |
|
|
Added a CheckType instruction for specific class membership testing. |
|
|
Fixed usage merging where names are defined below the current level in the |
|
|
Introduced a distinction between speculative attribute usage, which should |
|
|
Added comments for the attribute usage namespace attributes. |
|
|
Added attribute user node shelving and merging in order to affect nodes which |
|
|
Added recording of abandoned attribute usage specifically for attribute users |
|
|
Fixed ordering of statements which abandon attribute usage. |
|
|
Prevent re-entry into the attribute collection for an object. |
|
|
Introduced abandoned shelves of attribute usage which are used to provide |
|
|
Added control-flow branch abandonment for statements like break, continue, raise |
|
|
Made attribute deletion usage cause a warning during inspection, not an error, |
|
|
Distinguished between class and instance attributes when generating guards for |
|
|
Made InspectError a subclass of NodeProcessingError. |
|
|
Added optimisations note. |
|
|
Added tests of class attribute behaviour, including a test of "foreign" class |
|
|
Added information for interactive usage. |
|
|
Added a test of attribute access optimisation relevance where non-class, |
|
|
Changed the guard code generation to not attempt to generate guards for |
|
|
Added another reference. |
|
|
Introduced proper object visit tracking when collecting attributes. |
|
|
Added attribute usage registration for branches where more specific attribute |
|
|
Introduced a separate "unfinalisation" stage to avoid mixtures of old and new |
|
|
Regenerate the object and parameter tables after the program objects have been |
|
|
Introduced a test for classes as well as instances where specific types have |
|
|
Added missing namespace finalisation for classes. |
|
|
Reordered attribute/name traversal initialisation in order to avoid missing the |
|
|
Added "unfinalisation" of attributes in order to properly redefine structures. |
|
|
Introduced specific name usage so that attributes of classes and modules can be |
|
|
Added incomplete work to introduce attribute deletion and object table |
|
|
Changed the _attrnames annotation to map a number of names to attributes, thus |
|
|
Connected attribute usage sets to assignment nodes. |
|
|
Merged branches, retaining the namespace-based recording of attribute usage, |
|