|
⇤ ← Revision 1 as of 2004-01-27 17:37:20
Size: 27
Comment: Created by Aahz per GvR
|
Size: 1654
Comment: Aahz - moving stuff per GvR
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 1: | Line 1: |
| Describe CoreSprint here. | == PyConDC 2004 Core Sprint plan == Some ideas: * Complete the implementation of Generator Expressions * Add a module of statistical and reduction functions: stddev, average, nlargest, nsmallest, product, etc. * Examine two optimizations proposed by Neal Norwitz: + Add gotos from opcodes that set "why" since their jump target is known in advance + Add 2 custom op-codes for CALL_FUNCTION where the number of arguments is known in advance * Add 'diff3' and 'patch' to difflib * Unify doctest's two approaches to finding test cases. * Documentation sprint + People liked the examples in the docs for unittest, sets, and itertools. Apply that model to other parts of the docs. For instance, the email and logging packages would be a lot easier to use if there were a semi-comprehensive example section in the docs. * Python bug fixing sprint * PythonASTSprint -- finish the new compiler on the ast-branch -- this should be a top priority because the project has been open for so long. * Reduce python's start-up time. * Improve the byte code optimizer. Several new transformations can be added after checking for basic blocks and if the jump targets can be fixed up when the total number of bytes changes. Also, the line numbering has to be kept intact during the transformations. * Speed up list comprehensions by pre-allocating blocks of 100 list elements, writing using `PyList_SET_ITEM`, and adjusting the total size at the end of the comprehension. * Update, pare-down, and clean-up Demo and Tools/Scripts Back to SprintPlan2004 |
PyConDC 2004 Core Sprint plan
Some ideas:
- Complete the implementation of Generator Expressions
- Add a module of statistical and reduction functions: stddev, average, nlargest, nsmallest, product, etc.
- Examine two optimizations proposed by Neal Norwitz:
- + Add gotos from opcodes that set "why" since their jump target is known in advance + Add 2 custom op-codes for CALL_FUNCTION where the number of arguments is known in advance
- Add 'diff3' and 'patch' to difflib
- Unify doctest's two approaches to finding test cases.
- Documentation sprint
- + People liked the examples in the docs for unittest, sets, and itertools. Apply that model to other parts of the docs. For instance, the email and logging packages would be a lot easier to use if there were a semi-comprehensive example section in the docs.
- Python bug fixing sprint
- PythonASTSprint -- finish the new compiler on the ast-branch -- this should be a top priority because the project has been open for so long.
- Reduce python's start-up time.
- Improve the byte code optimizer. Several new transformations can be added after checking for basic blocks and if the jump targets can be fixed up when the total number of bytes changes. Also, the line numbering has to be kept intact during the transformations.
- Speed up list comprehensions by pre-allocating blocks of 100 list
elements, writing using PyList_SET_ITEM, and adjusting the total size at the end of the comprehension.
- Update, pare-down, and clean-up Demo and Tools/Scripts
Back to SprintPlan2004