[Gzz-commits] manuscripts/pointers article.rst

[Benja Fallenstein]

CVSROOT:        /cvsroot/gzz
Module name:    manuscripts
Branch:         
Changes by:     Benja Fallenstein <address@hidden>      03/11/01 15:47:09

Modified files:
        pointers       : article.rst 

Log message:
        abstract shorter, rhetoric into intro

CVSWeb URLs:
http://savannah.gnu.org/cgi-bin/viewcvs/gzz/manuscripts/pointers/article.rst.diff?tr1=1.53&tr2=1.54&r1=text&r2=text

Patches:
Index: manuscripts/pointers/article.rst
diff -u manuscripts/pointers/article.rst:1.53 
manuscripts/pointers/article.rst:1.54
--- manuscripts/pointers/article.rst:1.53       Sat Nov  1 15:39:07 2003
+++ manuscripts/pointers/article.rst    Sat Nov  1 15:47:09 2003
@@ -40,31 +40,26 @@
 Abstract
 ========
 
-If the Web worked like a file-sharing system, there would be
-no central point of failure for a Web page; it could be downloaded
-from any host that has a copy. This would save bandwidth
-and increase availability. However, if the Web worked like
-a file-sharing system, Web pages could never be updated.
+File-sharing systems save bandwidth and increase availability
+because there is no single point of failure for a file.
+This would be desirable on the Web.
+However, most current P2P systems have no facility 
+for updating documents, and those that do 
+require keeping the current version as state in the network.
+Thus, to keep a version of a Web page available, 
+it would not suffice to keep a copy on your computer.
 
-We present an updating system usable
+We present an versioning mechanism usable
 for a P2P Web as well as for file-sharing (e.g. corrected versions
 of Project Gutenberg books) and P2P-based software distribution
 (akin to Debian's ``apt-get``).
 
-Most current P2P systems have no facility 
-for updating documents. The systems that do 
-require state in the network 
-for remembering what the current version of a document is.
-Thus, to keep a version of a Web page available, 
-it does not suffice to keep a copy on your computer.
-
 Extending OceanStore's "heartbeats," we propose *pointer records*, 
 signed files containing a document's id, 
-the hash of its current version, and a timestamp.
-Pointer records, like other files, can be downloaded from any peer 
-that has a copy. We can find all available pointer records for a document 
-through the same mechanism we use to find files in the peer-to-peer network 
-(for example, a distributed hashtable). 
+its current version's hash, and a timestamp.
+We can find available pointer records for a document 
+through the same mechanism we use to find files by hash
+(for example, a DHT). 
 To find the newest version, we look for the most recent timestamp.
 This way, versions of a document 
 stay available as long as anybody keeps a copy.
@@ -83,6 +78,12 @@
 
 .. Introduce location-independent Web idea (citing
    benefits from hash-based addressing)---
+
+If the Web worked like a file-sharing system, there would be
+no central point of failure for a Web page; it could be downloaded
+from any host that has a copy. This would save bandwidth
+and increase availability. However, if the Web worked like
+a file-sharing system, Web pages could never be updated.
 
 .. Standing on the shoulders of giants: Example of Web links rotting away