bill@zycor.UUCP (bill) (11/14/89)
A Sys V question... If you set an alarm for, say, 3 seconds, then write to a named pipe when there is no process reading from it, the alarm never kills the read. Am I doing something wrong, or did I miss something in the FM? -- Bill Mahoney bill@zycor.UUCP The holiday season: time to wonder if the Salvation Army has tanks...
les@chinet.chi.il.us (Leslie Mikesell) (11/16/89)
In article <68@zycor.UUCP> bill@zycor.UUCP (bill) writes: >A Sys V question... >If you set an alarm for, say, 3 seconds, then write to a named pipe >when there is no process reading from it, the alarm never kills the >read. Am I doing something wrong, or did I miss something in the FM? Are you sure you actually got that far? The open() should have blocked waiting for a reading process unless you set O_NDELAY, in which case write() should also return immediately with an error when the pipe is full. Depending on the application, you may find it useful to have the writer open the FIFO in r/w mode (but never do a read()). This allows writing until the pipe buffer fills regardless of the reading process which sounds like what you are trying to do, but it prevents you from knowing if the reader is running. Les Mikesell les@chinet.chi.il.us