Author: toad
Date: 2008-09-25 22:55:43 +0000 (Thu, 25 Sep 2008)
New Revision: 22840
Modified:
trunk/freenet/src/freenet/node/Node.java
Log:
Explain why we pad by default
Modified: trunk/freenet/src/freenet/node/Node.java
===================================================================
--- trunk/freenet/src/freenet/node/Node.java 2008-09-25 22:19:46 UTC (rev
22839)
+++ trunk/freenet/src/freenet/node/Node.java 2008-09-25 22:55:43 UTC (rev
22840)
@@ -1600,6 +1600,16 @@
maxStoreKeys = maxTotalKeys / 2;
maxCacheKeys = maxTotalKeys - maxStoreKeys;
+ /*
+ * On Windows, setting the file length normally involves
writing lots of zeros.
+ * So it's an uninterruptible system call that takes a loooong
time. On OS/X,
+ * presumably the same is true. If the RNG is fast enough, this
means that
+ * setting the length and writing random data take exactly the
same amount
+ * of time. On most versions of Unix, holes can be created.
However on all
+ * systems, predictable disk usage is a good thing. So lets
turn it on by
+ * default for now, on all systems. The datastore can be read
but mostly not
+ * written while the random data is being written.
+ */
nodeConfig.register("storePreallocate", true, sortOrder++,
true, false, "Node.storePreallocate", "Node.storePreallocateLong",
new BooleanCallback() {
@Override
