<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000066">
<div class="moz-cite-prefix">On 6/6/12 12:36 PM, Joe Armstrong
wrote:<br>
</div>
<blockquote
cite="mid:CAANBt-roaAXw4KyNKYdRrYwprkCbtJ_4bTHyeJpWssUCiCY5ww@mail.gmail.com"
type="cite">
<pre wrap="">I'm really impressed by the real-world haskell site (see for example
<a moz-do-not-send="true" class="moz-txt-link-freetext" href="http://book.realworldhaskell.org/read/functional-programming.html">http://book.realworldhaskell.org/read/functional-programming.html</a>)</pre>
</blockquote>
<br>
Since you already mentioned that you were thinking of UUIDs, maybe
the way was to set up a global comment repository and provide
JavaScript links and alternatively a browser plugin that can link to
/any/ paragraph on any page to let comments appear after any
paragraph anywhere. The JS could be for sites to tie in, which
intent to participate. They would not have to change anything in
their structure as the JS would just inspect the DOM and add the 'n
comments' hooks. The browser plugin for all other.<br>
<br>
This annotated web could be complemented with the possibility to
also store links between any two paragraphs explicitly, beyond the
obvious option of writing URLs into the comment proper.<br>
<br>
To compensate for multiple copies or dynamic URLs, the text,
stripped of tags, could be hashed to arrive at a technical
reference.<br>
<br>
This would all only work in 70% of cases. Superficial changes would
break the hashes. But it could be an easy way to phase such efforts
in.<br>
<br>
Henning<br>
</body>
</html>