# HG changeset patch
# User Ethan Blanton <elb@pidgin.im>
# Date 1069869544 0
# Node ID 834570cb8d8629c973129adeb16a3b639c10277e
# Parent  9008b5be42755fefe4c7d306829d90ada0963ee0
[gaim-migrate @ 8286]
datallah seems to think that just compiling isn't enough, and that
it should also be correct.  Picky, picky...

committer: Tailor Script <tailor@pidgin.im>

diff -r 9008b5be4275 -r 834570cb8d86 src/gtkconn.c
--- a/src/gtkconn.c	Wed Nov 26 17:28:13 2003 +0000
+++ b/src/gtkconn.c	Wed Nov 26 17:59:04 2003 +0000
@@ -314,24 +314,33 @@
 		gtk_tree_model_get_value(model, &iter, 4, &val);
 		account = g_value_get_pointer(&val);
 		g_value_unset(&val);
-		gaim_account_connect(account);
-		/* remove all disconnections of the account reconnected */
+		/* remove all disconnections of the account to be reconnected */
 		if (gtk_tree_model_get_iter_first(model, &iter)) {
+			GList *l_del = NULL;
 			GaimAccount *account2 = NULL;
-			gboolean alreadyIterated = FALSE;
 			do {
-				alreadyIterated = FALSE;
 				gtk_tree_model_get_value(model, &iter, 4, &val);
 				account2 = g_value_get_pointer(&val);
 				g_value_unset(&val);
 				if (account2 == account) {
-					gtk_list_store_remove(GTK_LIST_STORE(model), &iter);
-					alreadyIterated = TRUE;
+					GtkTreePath *path = gtk_tree_model_get_path(model, &iter);
+					GtkTreeRowReference* del_row = gtk_tree_row_reference_new(model, path);
+					l_del = g_list_append(l_del, del_row);
 				}
 				g_value_unset(&val);
-			} while (alreadyIterated || gtk_tree_model_iter_next(model, &iter));
+			} while (gtk_tree_model_iter_next(model, &iter));
+
+			while (l_del != NULL) {
+				GtkTreeRowReference* del_row = l_del->data;
+				GtkTreePath *path = gtk_tree_row_reference_get_path(del_row);
+				if (gtk_tree_model_get_iter(model, &iter, path))
+					gtk_list_store_remove(GTK_LIST_STORE(model), &iter);
+				l_del = l_del->next;
+			}
 		}
 
+		gaim_account_connect(account);
+
 		if (!gtk_tree_model_get_iter_first(model, &iter))
 			disconnect_window_hide();
 		else {