ryanpoulos / labview-memory Goto Github PK

check with R&D to see what name should be.

Current the swap method no-ops when the buffer sizes are mismatched. Ideally we should return an error to prevent silent failures. Rememeber: Fail early! Fail often!

Currently there is nothing that prevents me from double disposing a memory block. We should add a pointer check to the (safe) API before disposing memory.

Currently, the Pointer class' Read and Write methods for 1D arrays only checks the Pointer to make sure it has a large enough size for the transfer. This is good, but LabVIEW can still crash if the provided buffer array doesn't have enough (count) elements. So it should check both conditions (Pointer size and array size).

Currently there is no synchronization that prevents me from reading/writing/disposing a buffer at the same time from multiple threads. This could lead to race conditions where we crash or get corrupt data based on timing of the system. We should add some synchronization primitives to have atomic read/write/dispose methods.

We need test coverage on all (public) methods. Use a testing framework, like VI tester, to write unit tests. I have forked your code and added a few basic tests for an example.
https://github.com/csjall/LabVIEW-Memory-Manager

To improve readability, make an explicit method to convert counts to bytes.

All No Check 1D Write functions shouldn't be offset out calculated from value of offset in + count/size?

Then when you want daisy-chain more writes to one pointer then you don't have to calculate offsets for consecutive writes by yourself.

Memory Manager\Classes\Pointer\Public\Write (No Check)\Write (No Check 1D xxx).vi