Patrick Chanezon's weblog

Reposting: Blogging mailing lists: an idea about the implementation

Reposting and old entry in order to be able to link to it, because my Radio weblog is going down in a month and I haven't had time to migrate the data yet.

Blogging Mailing Lists. I wish there were better tools for blogging mailing lists, so that these two types of conversations could cross over more often. [Meerkat: An Open Wire Service]

Excellent idea.

I think userland has that feature, send a mail to your blog, but what would be truly valuable would be to link automatically to the archived web version of the mail, in its thread context.

In order to do this we would need 2 things:

a standard WSDL for web archive of mailing list searching, something like MailArchiveSearch.wsdl:

• It could allow different modes of searching: date + author, or full text
• it returns the url of the mail in the archive

An index somewhere, that could be decentralized using RSS, that would include the following informations:

• mailing list name: the key for lookup
• mailing list archive web site
• url of the entry point for the service that implements MailArchiveSearch.wsdl for that site (it could be implemented at a remote site, for example one that would aggregate search for different archives. a Google front end could do most of the archives I guess)

Then when you forward the mail you are interested in, with your comments, to your favorite blogging tool, it would:

• extract the mailing list name
• lookup the mailing list search service, using an RSS aggregator like Merkaat
• invoke the search service
• if something is found, add a permalink to your post, to the archived email
• else, if no search service is found or if the mail is not found, just add your blog entry as is

I may have a stab at implementing this for Radio, with the jakarta mailing lists, and maybe Google API... if I can free up some time for a personal project :-)

...Obviously I did not have much amount of free time in the past 2 years :-)

Re: Russel Beattie's Web Service Tools entry

In Web Service Tools Russ shares some thoughts about Web Services as Unix-like Software Tools.

This entry rung a bell for me for 2 reasons:

• It's been 2 months that I've been ejoying the delightful reading of Eric Raymond's The Art of Unix Programming, where he distills what makes the UNIX software design philosophy with many detailed examples. I'm only at page 300, on 510. I haven't enjoyed a software related book so much since the GoF Design Patterns book. While reading I've been thinking about how web services implemented using the REST style exhibited the same desirable compositional characteristics as UNIX command lines using pipes.


• Recently I have started playing with such a REST style web service, the Amazon Web Services, which can be used using either SOAP or REST. My little experiment consisted in replacing the stylesheet used by Amazon to generate RSS feeds by an alternate stylesheet which adds information to the feed, here the list of weblogs that talk about that feed. I get the list of weblogs from another REST style web service provided by AllConsuming. The Amazon RSS feeds themselves are generated using the equivalent of a pipe: parameters on the url generate the request for XML data from Amazon, and one parameter redirects that XML to the Amazon xslt service for styling in RSS.
  I could not use Amazon's xslt service because I use a document() function in my stylesheet that fetches some additional XML from AllConsuming. Thus I used the W3C XSLT service to apply my stylesheet: like in UNIX where you can easily change one filter in a chain of commands bound using pipes, I just had to change the url for the xsl service.
  I need to go one step further: here I cheated by using the Amazon XML results directly to do my transformation. I did this for efficiency reasons, and also because I was too lazy to parse the urls to get the ASIN out of them. But my logical next experiment is to take the feeds directly from Amazon, and then apply my stylesheet to add the AllConsuming data: in the UNIX metaphor, this would truly be like piping Amazon's RSS to another filter.
  Will do that when I find some time.
  My experiment is explained in Enriching Amazon RSS feeds with AllConsuming metadata: take 2
Posted by chanezon at 05:00 PM | Comments (0) | TrackBack

Blogtalk 2.0 proposal: Blogging at Sun 2002-2004: a case study of weblog adoption in a technology company

Blogtalk 2.0 call for paper's deadline is today.
According to the site:

The conference will be held in Vienna, Austria in July 2004 (more details coming up pretty soon)

BlogTalk 2.0 is designed to initiate a dialog between bloggers, developers, researchers and others who share, enjoy and analyse the benefits of blogging. The focus is on weblogs as an expression of a culture based on the exchange of information, ideas and knowledge. In the spirit of the free exchange of opinions, links and thoughts we wish to engage a wide range of participants from the blogosphere in this discourse.

Last year I proposed a presentation Weblogs and Portals: can David and Goliath collaborate ? but it wasn't accepted.
I guess it was a bit too geeky and not specific enough for the audience :-)

I came to the conference with my Sun pal Alejandro "portlet" Abdelnur and had a lot of fun and very interesting discussions with many participants and presenters (I'm the guy on the right on Heiko's pictures, Alejandro the second from the left :-).

This year, after one more year experimenting with weblogs, and Sun's recent interest in weblogs and RSS and push to put some serious resources on that topic, I hope I have a more relevant presentation for the blogtalk audience: Blogging at Sun 2002-2004: a case study of weblog adoption in a technology company
I hope they'll accept it this time. Comments are welcome.

Here it is:
Blogging at Sun 2002-2004: a case study of weblog adoption in a technology company

Author's last name and first name: Chanezon Patrick
Affiliations (country and name of institution or company): France, Sun Microsystems
URL: http://www.chanezon.com/pat/weblog/
Title of the paper: Blogging at Sun 2002-2004: a case study of weblog adoption in a technology company

Proposal
In this paper I will use the word weblogs for weblogs and related technologies that started to get some traction in 2002. This includes weblogs, all falvors of RSS and Atom, RSS online aggregators and desktop readers, Wikis, and some Semantic Web applications such as FOAF.

From its inception as a UNIX workstation provider to the networked systems provider it is now, Sun always had a mixed culture in hardware and software design, the software side being very much influenced by the UNIX culture.
From 2002, weblogs begun to have an adoption curve similar to the beginning of the web, in a mostly bottom up fashion, which was very well demonstrated by numerous presentations at blogtalk 1.0 in 2003.

For most companies weblog related technologies can be used in two ways: internally for a more efficient collaboration; externally to create and foster communities around its technologies, and offer a more human voice and image than classic Public Relations.
For companies producing software or software based services such as Sun, IBM, Microsoft, BEA, Oracle, Macromedia, Google or SixApart there is another dimension to weblog technologies, which is creating products or services implementing these technologies.

This case study will describe Sun’s adoption of weblogs from 2002 to today, both for internal and external use, and from a product and strategy perspective. Then this history will be analysed according to different factors: company history, organizational structure, software culture, economic context, technical environment. Some of these factors are a mix of top-down and bottom-up approaches, the prevalence of UNIX culture over web culture, with the high use of email as a collaboration tool, the roots of the company as an engineers company, the economic situation of Sun and its software strategy, oriented towards turnkey scalable network based systems and java.

A comparison with other companies adoption history will be tried, based on informations available to the public. Last an attempt will be made at determining some general factors from the previous analysis that influenced weblog adoption patterns by technology companies, as well as at predicting what may be in store for the close future.

I am a only software developer with an interest in history and sociology, so this case study won’t pretend to be at the accademic-grade level of quality. However I think the personal and company experience described here can be leveraged in the future by true sociologists and historians for more complete analysis of the topic. Moreover I think the questions I raise about the impact of various sociological, economical and cultural factors on weblog adoption are important to consider for enterprises wishing to adopt weblogs today. Last the discussion about technology companies strategies regarding these technologies can be a good starting point to look at this area’s future.

Enriching Amazon RSS feeds with AllConsuming metadata: take 2

In yesterday's entry I created a stylesheet to enrich Amazon RSS feeds with AllConsuming metadata, but could not give you a url to test the feed because Amazon xslt service doesn't let you fetch external urls.
I used an ant local build file to generate the feed on my machine.

Today I've tested it using W3C's excellent XSLT Service.
It's not to be used heavily but it's OK for a few tests.

The xml data is a REST query to Amazon API for the latest books in the Literature & Fiction category.
This is the same query that is used in the corresponding Amazon RSS feed.
Amazon Literature & Fiction XML Data

Instead of feeding it the Amazon xslt stylesheet I feed it my modified stylesheet.
http://www.chanezon.com/pat/amazon/xml-rss091-with-allconsuming.xsl

The full url to use for the xslt service is:
Amazon Literature & Fiction RSS feed with added AllConsuming metadata

I tested it in NetNewsWire and it works fine, but not all the time: the stylesheet is very greedy in server resources and takes a long time to get interpreted. My first trial gave me a timeout.

The whole promise of standardizing feeds on RSS is to be able to filter these feeds and enrich them using standard tools.
Adding AllConsuming data to the Amazon RSS seems to me like a perfect example of the power of this approach.
Next step is to look at the RSS 1.0 extension modules and determine in which RDF schema these metadata should live: adding them in escaped HTML form to the description element is nice for existing RSS readers, but in the next few months I expect readers to expand their capabilities in the semantic tag interpretation.

Enriching Amazon RSS feeds

A few days ago Amazon published some RSS feeds for some of their categories.
This is a simple application of their great Web Services APIs, released since 2 years already, and of their XSLT service.

These feeds were possible before but you had to craft your own stylesheet.
Now that these feeds have a page of their own, regular people will be able to point their RSS agrregator to these using the orange XML buttons instead of having to create their own xsl first :-)

I've started looking at the Amazon APIs in the past 2 weeks: I had played with Google's APIs last year, but not with Amazon's yet.
They cover much more functionalities.

This new RSS feature prodded me to go a bit further.
I've enriched their default stylesheet a bit to enrich the Amazon feeds with informations taken from the AllConsuming seb site, such as the number of weblogs which mentioned the book in the past 60 days, and the urls to these weblogs.
These informations are already present in AllConsuming feeds, but now you can add them directly to the amazon feeds.
This is accomplished thanks to the REST interface that AllConsuming offers to poll the data out of its site.

The stylesheet polls AllConsuming for each isbn and formats the returned data in HTML in the description entry.
I can then read that in my Desktop RSS NewsReader, NetNewsWire.

This is a first draft and it needs more work, but I found it to be a good demo for the benefits XML, REST, SOA, WS and other acronyms you may have heard of but don't know what use they could be for you.

The stylesheet is at
http://www.chanezon.com/pat/amazon/xml-rss091-with-allconsuming.xsl
It just adds a few lines to the original
http://xml.amazon.com/xsl/xml-rss091.xsl

It works when invoked locally on my machine (I launch it from ant), but doesn't work when using Amazon's xslt service.
http://xml-na.amznxslt.com/onca/xml3?mode=books&bcm=Books%3A%20Literature%20%26%20Fiction&t=webservices-20&dev-t=amznRss&type=lite&page=1&ct=text/xml&sort=+salesrank&f=http://www.chanezon.com/pat/amazon/xml-rss091-with-allconsuming.xsl&BrowseNodeSearch=17

I think it is because they must forbid accessing external urls using the document() function in your stylesheet, which I can understand.
I need to host a simple xsl service on my box... maybe tomorrow.

Note: in order to get some HTML in the description that displays nicely in NetNewsWire I had to escape all markup in my xslt, which reminded me of Norman Walsh' excellent thread "escaped markup considered harmful".
Norm was so right !

Google Toolbar on Mozilla, and thoughts about comment spamming and trust

Syndic8's Jeff Barr half swicthed to Mozilla but misses Google and Alexa Toolbars.
The mozdev project Googlebar, provides a pretty good equivalent of the Google Toolbar functionality.
I use it since I switched to a Mac and am happy with it.
For Alexa I did not find anything: maybe the Googlebar team should enhance it to add Alexa's features.

PS: I initially wanted to add a comment on Jeff's weblog but he closed comments off. I'll try to Trackback it with this entry. This is the sad reality of spam coming to weblogs through comment spam.
Erik Benson's proposed Talkrnet proposal would be a good first step to solve this issue without blocking communication. Using social software to determine who's trusted to comment is a good ad-hoc practical first step. The next step will be a standardization of the interface to such a trust service: managing digital identity and trust and the next big problem to be solved.