compilation cache: methods have a chance to work

This commit is contained in:
Araq 2011-10-23 11:24:52 +02:00
commit bd1cb9e77b
7 changed files with 100 additions and 35 deletions

View file

@ -1,25 +1,34 @@
Version 0.8.14
==============
- optimize unused constants away
- optimize unused constants away (affected by HLO)
- 'let x = y'
- fix actors.nim
- make threadvar efficient again on linux after testing
- test the sort implementation again
- document & test splicing; don't forget to test negative indexes
- implement lib/pure/memfiles properly
- eval context is per module; this way modularity is kept; global id generation
macro can still be done once macros support basic IO (store current id in
some file)
incremental compilation
-----------------------
- adapt thread var implementation to care about the new merge operation
- write test cases: needs test script support
- test type converters
- test G, A, B example from the documentation
- test thread var
- test method generation
- test method generation; could work, needs ROD support though
- test init sections
- test DLL interfacing!
- hallo.rod is missing initial statements: feature or bug?
- automate tests:
- test basic recompilation scheme
- test type converters
- fix remaining bugs
@ -57,6 +66,9 @@ Bugs
result = forward(x)
- bug: DLL generation is broken
- bug: stress testing basic method example (eval) without ``-d:relase`` leaks
memory; good way to figure out how a fixed amount of stack can hold
an arbitrary number of GC roots!
version 0.9.XX