(Subject: changed to be more specific.) On Sun, 10 Nov 2019 08:00:24 -0600, Pierre Fichaud wrote:
>I want to process Unix files in a sub-task. The files are being written into >by applications in other address spaces. I open the file (BPX1OPN). I need to >read all the records. errno EAGAIN seems to say that there are no new records >in the file. So I would WAIT and retry. > I have done this by issuing a blocking read for a single byte to perform the wait rather than polling, then non-blocking reads for remaining available data. >... Does this make sense ? I also want to know if the file is still being >written into. BPX1FST (file status) will return an area. Does the field >ST_NLINK tell me how many "OPENs" have been done ? If I'm the only one >connected to the file, would ST_NLINK have a value of 1? > 2, one for the directory link, one for you. This may be unreliable if o Multiple directory links exist or another task unlinks the file. o fork(), spawn(), or dup() may affect the link count. o A producer task neglects to close the file promptly. I have done this only with unnamed pipes. Behavior of regular files may be different. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
