async has become a reserved word. I suggest that sanlock starts supporting async_ arg as well, as does ovirtsdk4 https://gerrit.ovirt.org/gitweb?p=ovirt-engine-sdk.git;a=commitdiff;h=7f75f4786c82b3df192257a6c14a19a7dfc1c527
async
sanlock
async_
Metadata Update from @nsoffer: - Issue assigned to nsoffer
In sanlock 3.8, first version supporting python 3, async=False will be replace by wait=True. The python 2 version will support both async and wait flags for backward compatibility.
In python 2 and python 3 we will support: - wait=True - wait=False
In python 2 we will allow also: - async=True - async=False
If both async= and wait= arguments are specified, raise RuntimeError.
@nsoffer Maybe we should use the term "block" instead of "wait", then block = True would be like async = False, but the intention is more clear IMO.
sanlock already chose wait=, for example:
inq_lockspace(lockspace, host_id, path, offset=0, wait=False)
Vdsm went with wait= as replacement for async. The concept of wait is very popular in the standard library, and its more general and clear.
okay. we shall wait.
github patches, i believe resolution there eliminates the need to check for coexistance of "wait" and "async" in Py2: https://github.com/nirs/sanlock/pull/14/commits/83bc4e375db06b383c8fab7d5977b7631fc37310 https://github.com/nirs/sanlock/pull/14/commits/e7230b58a8d5f379b891788255231167b2d28442
This was resolved in sanlock 3.8/0.
Metadata Update from @nsoffer: - Issue status updated to: Closed (was: Open)
Login to comment on this ticket.