Todo List

Member cd::CDHandle::GetErrorProbability (SysErrT error_type, uint error_num,)

Decide on rate vs. number+probability

Member cd::CDHandle::RegisterRecovery (uint error_name_mask, uint error_loc_mask, RecoverObject *recover_object=0)

Does registering recovery also imply turning on detection? Or is that done purely through RequireErrorProbability()?

Member cd::CDHandle::RequireErrorProbability (SysErrT error_type, uint error_num, float probability, bool fail_over=true)

Decide on rate vs. number+probability

Member cd::CDHandle::SetPGASUsage (void *data_ptr, uint64_t len, PGASUsageT region_type=kShared)

Do we want to expose explicit logging functions?

Class cd::CDNameT

Decide on whether to represent a CD name as (level, num) or as typedef std::vector<uint> CDNameT; to represent the path through the tree branches.

Member cd::SysErrLocT

is SysErrLocT comprehensive enough for portability?

Member cd::SysErrNameT

Is SysErrNameT comprehensive enough for portability?

segv (segmentation violations) can be used as proxy for soft memory errors using the existing kernel infrastructure

Group cd_error_probability

What about specifying leniant communication-related errors for relaxed-CDs context?

Member CDInternalPtr::InternalReexecute ()

Discuss other aspects of reexecution (e.g., logging). [FIXME] Discuss other aspects of reexecution (e.g., logging).

page Containment Domains

Write a more significant introduction and put sections for what are currently modules

Class RecoverObject

Write some example for custom recovery (see GVR interpolation example, although they do it between versions).