+27Jan02=======================================================================
+1. make save work in cron context
+2. try to get the C client to handle closes on its standard input better
+3. put some validation in cluster.pl to check that people have set it up
+handled.
+22Jan02=======================================================================
+1. limit spot dupe checking to first 12 characters
+13Jan02=======================================================================
+1. changed the backstop termination alarm routing in the C client so that
+it flushes the input and output buffer and then commits suicide.
+12Jan02=======================================================================
+1. put setsockopts into eval block to allow some of them to fail
+11Jan02=======================================================================
+1. Make M$ TCP connx blocking again, sigh...
+2. backed out ALL the M$ speedup and make nicer changes :-(
+3. cope with data coming in during the disconnection process. Don't worry
+Earth people, this will take no longer than 15 of your Earth seconds...
10Jan02=======================================================================
1. set SO_LINGER = 0 to try to prevent the system sending data on a closing
-socket. This MAY help some of the hanging problems.
+socket. This MAY help some of the hanging problems (but I am now beginning
+to doubt this).
2. do the same on the C client (which means DON'T FORGET to make it).
+3. set no nagling on the output from the node.
+4. set KEEPALIVE on TCP connections for both the node and C Client (you will
+have to make the C client again) and you will need to restart.
+5. It appears that it IS possible to set non blocking on M$ you just need to
+know how.
+6. Fix problem with load/cmd_cache core dumping the system
08Jan02=======================================================================
1. altered sh/qrz to point to the new server
2. alter the character set handling a bit to make it better for spanish
projects / spider.git / blobdiff