[Archipelago-submits] [232] trunk/archipelago/TODO: updated TODO

[nobody at rubyforge.org]

Revision: 232
Author:   zond
Date:     2007-03-07 07:18:32 -0500 (Wed, 07 Mar 2007)

Log Message:
-----------
updated TODO

Modified Paths:
--------------
    trunk/archipelago/TODO

Modified: trunk/archipelago/TODO
===================================================================
--- trunk/archipelago/TODO	2007-03-07 11:46:27 UTC (rev 231)
+++ trunk/archipelago/TODO	2007-03-07 12:18:32 UTC (rev 232)
@@ -1,19 +1,4 @@
 
- * Create a failover/redundancy framework
-   * For example: Create a new HashishProvider with built in redundancy, 
-     for example using the Chord project: http://pdos.csail.mit.edu/chord/
-   * Or: Create migration methods that move objects between Chests opon
-     startup and shutdown, and make them keep backups at each others
-     persistence backends.
-   * Or: Create something that stores data the way Chord does (with erasure
-     codes) but doesnt use the same look up mechanism.
-     * Problem: We still have to implement the entire maintenance protocol
-       of Chord (continously checking if our data is safely replicated across
-       the network, continously checking that our data belong with us)
-
- * Replace Raider with some well known near-optimal erasure code, for example
-   Online Codes: http://en.wikipedia.org/wiki/Online_codes
-
  * Make Chest aware about whether transactions have affected it 'for real' ie 
    check whether the instance before the call differs from the instance after 
    the call. Preferably without incurring performance lossage.
@@ -30,3 +15,5 @@
  * Create a memcached-client that uses Disco instance to find all memcached instances
    in the network and distribute requests among them.
 
+ * Create a file server services that handles really big files within transactions
+   and allows the POSTing and GETing of them via HTTP.