chromatic wrote:
Hi all,
In a recent discussion with Chip and Leo, the idea came up to ask for a
list of very specific TODO items -- specifically things that should work
but don't.
Some of the items in this thread are TODOs, some bugs, and some could be
noops. I'll try to classify items in this thread.
I know Autrijus at YAPC::NA mentioned that the lexical implementation
didn't really work.
Lexicals are subject to an upcoming design document. While the
implementation basically works, it doesn't really fit into HLLs
expectations. The static, compiler-visible lexicals of one sub can be
"flattened" and will be directly mapped to Parrot registers. The current
approach of fully dynamic lexicals can be minimized, depending on the
strict-ness of the HLL.
I think that nested ManagedStruct and UnManagedStruct PMCs share state
between them, so you can't have two independent data structures -- but
I'm not completely sure.
Not AFAIK. A *Struct PMC holds the names and offsets of one C structure.
Nested structures aren't really different here, there is just another
*Struct PMC for the nested structure. The ManagedStruct owns
additionally the memory occupied by the structure, while the
UnManagedStruct points to memory external to Parrot.
-- c
leo
