Closed jishnub closed 4 years ago
Some of the commits are relics of a previous PR that has already been merged. Only the last 3 are new.
Tests appear to fail on Mac OS because of BinaryProvider. Maybe need to wait for a version bump from them.
Not sure why BSD tests are failing
Can you please rebase on master
? I made some changes to CI
Thanks, tests pass now
Hope I've not screwed up the rebase by squashing the commits
I hope you don't mind I have force-pushed to your branch, but the history was messed up.
Opening a pull request from a branch that has the same name as the target one is considered an antipatter and GitHub shows you a big red warning when you try to do so. The problem is exactly what you faced: when you have to sync with upstream the history can become a mess if you don't know how to do it. Help yourself, create a branch with a different name the next time :wink:
Thanks for your help, for some reason I seem to have missed the warning. Is this ok to merge now?
This PR lets us read and write complex arrays that are reinterpreted as floating point ones (addresses #129). Eg:
The
reinterpret
wrapper is cheap and allocates minimal memory, so this avoids the need allocate a separate real array. Contiguous views work as expected as well.Additionally, this also changes the compatibility bounds on dependencies from inequalities to caret ones (#130). This is more conservative and keeping in line with the julia semver convention.