> right. there are many things that we talked about with Russell (a long > time ago) that are still to be done. And I'm pleased to see that > you're working on it ;-)
The most urgent problem, is probably blocking on connect() that was troubling for a while in the 'netxml-ups' driver, but turns out to be just as big a problem in the upsclient library. Until recently, we have worked around blocking on read() and write() in some clients (most notably, 'upsmon') by prepending alarm() calls to the library functions. Now that the read/write functions to the library no longer use blocking I/O, these can (and will) be removed. Surprisingly, we never used this mechanism on upscli_connect(), which can/will block just like the read/write actions did. Until recently, I wasn't even aware that this might be (and actually is) a problem. Should we attempt to solve this in nut-2.2.2 or (preferably) leave this for nut-2.4.0 (so that we have a bit more time to see if the changes are portable)? Best regards, Arjen _______________________________________________ Nut-upsdev mailing list [email protected] http://lists.alioth.debian.org/mailman/listinfo/nut-upsdev
