Re: Is flock necessary?

flock() may not be required - but I'm pretty sure it's damn sensible. As for using mySQL, well, these routines are part of a simple file-based cache system (specifically in my case used for caching mySQL data) Jtjohnston <taylorjo@collegesherbrooke.qc.ca> wrote in message news:3C86F290.550B957B@collegesherbrooke.qc.ca... > Group: Is flock even necessary? > Allan: You look like you have Perled. It was necessary for Perl. > Have you thought of MySQL instead? > > > Alan McFarlane wrote: > > > I've just written a couple of simple routines which read an write small text > > files to the server: > > > > function readData( $filename, &$data ) > > { > > if (!$fd = @fopen($filename, "r")) > > { > > return false; > > } > > > > flock($fd, LOCK_SH); > > $data = fread($fd, filesize($filename)); > > fclose($fd); > > > > return true; > > } > > > > function writeData( $filename, $data ) > > { > > if (!$fd = @fopen($filename, "w")) > > { > > return false; > > } > > > > flock($fd, LOCK_EX); > > fwrite($fd, $data); > > fclose($fd); > > > > return true; > > } > > > > Now, the question is... How much time elapses between the fopen() statement > > and the flock() statements? > > > > Looking at the code, it would appear that two separate threads may call > > writeData() almost simultaneously giving the following execution: > > > > assume $filename = "sample.txt"; > > > > Thread #1 - $fd = fopen($filename, "w"); // succeeds > > Thread #2 - $fd = fopen($filename, "w"); // succeeds (I think) > > Thread #1 - flock($fd, LOCK_EX); // Does this succeed or fail? > > Thread #2 - flock($fd, LOCK_EX); // fails (I think) > > -- > John Taylor-Johnston > -------------------------------------------------------------------------- --- > ' ' ' Collège de Sherbrooke: > ô¿ô http://www.collegesherbrooke.qc.ca/languesmodernes/ > - Université de Sherbrooke: > http://compcanlit.ca/ > 819-569-2064 > >