couchdb-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From fdman...@apache.org
Subject svn commit: r1134943 - /couchdb/trunk/src/couchdb/couch_replicator.erl
Date Sun, 12 Jun 2011 16:20:11 GMT
Author: fdmanana
Date: Sun Jun 12 16:20:11 2011
New Revision: 1134943

URL: http://svn.apache.org/viewvc?rev=1134943&view=rev
Log:
Update comment to mention a supervisor change in R14B03

Modified:
    couchdb/trunk/src/couchdb/couch_replicator.erl

Modified: couchdb/trunk/src/couchdb/couch_replicator.erl
URL: http://svn.apache.org/viewvc/couchdb/trunk/src/couchdb/couch_replicator.erl?rev=1134943&r1=1134942&r2=1134943&view=diff
==============================================================================
--- couchdb/trunk/src/couchdb/couch_replicator.erl (original)
+++ couchdb/trunk/src/couchdb/couch_replicator.erl Sun Jun 12 16:20:11 2011
@@ -259,18 +259,15 @@ do_init(#rep{options = Options, id = {Ba
          io_lib:format("`~s`: `~s` -> `~s`",
             [BaseId ++ Ext, SourceName, TargetName]), "Starting"),
 
-    % Restarting a supervised child implies that the original arguments
+    % Until OTP R14B03:
+    %
+    % Restarting a temporary supervised child implies that the original arguments
     % (#rep{} record) specified in the MFA component of the supervisor
     % child spec will always be used whenever the child is restarted.
     % This implies the same replication performance tunning parameters will
-    % always be used. See the following threads for details:
-    %
-    % http://www.erlang.org/cgi-bin/ezmlm-cgi?3:sss:1772:201012:kihiniifeclgnpodlipd#b
-    % http://erlang.2086793.n4.nabble.com/PATCH-supervisor-atomically-delete-child-spec-when-child-terminates-td3226098.html
-    %
-    % The current solution is to delete the child spec (see cancel_replication/1)
-    % and then start the replication again, but this is unfortunately not immune
-    % to race conditions.
+    % always be used. The solution is to delete the child spec (see
+    % cancel_replication/1) and then start the replication again, but this is
+    % unfortunately not immune to race conditions.
 
     ?LOG_INFO("Replication `~p` is using:~n"
         "~c~p worker processes~n"



Mime
View raw message