Attention is currently required from: pespin, daniel.
laforge has posted comments on this change. ( https://gerrit.osmocom.org/c/libosmocore/+/32759 )
Change subject: [WIP] osmo_io: Support detecting connect ......................................................................
Patch Set 4: Code-Review+1
(2 comments)
Commit Message:
https://gerrit.osmocom.org/c/libosmocore/+/32759/comment/61c2bd05_179847cb PS4, Line 9: non blocking connect() which is signalled by : marking the fd writable. "non-blocking connect(), which as per definition of the socket API is signaled from OS to user by marking the file descriptor writable"?
File src/core/osmo_io_poll.c:
https://gerrit.osmocom.org/c/libosmocore/+/32759/comment/0accc9cd_65aa84ae PS2, Line 118: if (osmo_iofd_txqueue_len(iofd) == 0)
AFAIU it's the job of the user of this API to control this kind of enabled/disabled? […]
in general, with osmo_io, the question is "why would you want to bother the user with XYZ". If the answer to that question is "no", then it should be handled by the osmo_io provider (library) and not the user.
The user shouldn't know about low-level details like this. the user shouldn't know low-level implementation details of specific backend implementations (like if there's a libosmocore write-queue at all).
The user wants to write data, and get notified once it is written, or once it has failed to write. Anything else is an implementation detail.