Conversation
|
@jonnew while working on this I ended deviating on the proposals in #34 and I ended with something like this: Lines 75 to 88 in e95f1d1 Line 106 in e95f1d1 The reason is that individual function signatures were having too many parameters, which made them confusing. The main advantage of this new pattern, which is common in many POSIX syscalls, is that now an user can create a batch of arbitrary reads and writes to any device, which is something allowed by the hardware spec. The disadvantage is that the usage is less obvious. A user needs to create an array of To make this process easier, I also created some helpers for what's gonna be the most common batch operations: Lines 117 to 137 in e95f1d1 We can discuss different patterns, though. That is the main purpose of this PR, to discuss implementation details. (note that the old |
This is a draft PR to discuss the changes related to updating the library to the ONIv2 spec
The first commits contain an updated register interface allowing for batch register operations.
Currently only the ft600 driver translator is compatible with the new API