Saturday Apr 26, 2008

Microformats go mainstream

I talked about Microformats in a post last year on web20expo. It appears that the technology is now going main stream. I attended a workshop on Web2.0 Best Practices
at the Web20 Expo this week in which the speaker, Niall Kennedy
expounded on th advantages of using microformats. He said he's seen a
significant growth in traffic on his site since he started doing so since search engine results show direct links to pages on his site.
 Yahoo is adding microformats to many of their properties. The yahoo event site
already has them. This is exciting since microformats are a bridge to
the semantic web, which we've been talking about for several years now.
However, the talk has never seemed to materialize into anything
concrete. Meanwhile, the web2.0 world has decided to do things their
own way.

A classic example is tagging. While the semantic
folks talk about taxonomies and ontologies, the web guys invented
folksonomies (aka tagging). Tagging has allowed users and sites to
group stuff together, attaching semantic meaning to their data. Tag
clouds have worked fairly well and sites like flickr are extending the
concept by automatically creating even more tags ! The problem with
tags of course is that a word can have several meanings and it's not
easy to figure out which exact interpretation should be used. This
problem is what RDF solves nicely, but more on that later.

are better than tags in the sense that they have a more rigid format
and as such provide better semantics, although not perfect. Let's look
at an example:

<div class="vevent"><br>  <span class="summary">JavaOne Conference</span>: <br>  <span class="description">The premier java conference</span><br>  <p><a class="url"><a href=""></a><br>  <p><abbr class="dtstart" title="2008-05-06">May 6</abbr>-<br>  <abbr class="dtend" title="2008-05-09">9</abbr>,<br> at the <span class="location">Moscone Center, San Francisco, CA</span><br> </div>
which will display as :

JavaOne Conference:
The premier java conference

May 6-
at the Moscone Center, San Francisco, CA

advantage of such a format is that it clearly specifies various
properties associated with the event: summary, description, url, start
and end dates, location etc. However, it can still be ambiguous since
it uses literals for many properites e.g. the location. If someone
specified the location simply as "San Francisco", it could mean any of
27 different San Francisco's.

If we take this formalizing a step further, we reach the world of RDF.
Here every entry is specified as a tuple of the form:
<subject><predicate><object> using URIs to represent
the objects in an unambiguous manner. Without going into the syntactic
details, we could specify a location to be defined in the standard
format of: number, street, city, state, country, zip. This provides an
object with identity, the property that uniquely identifies it.

I'll talk more about RDF and semantic web in another post.

Friday Feb 29, 2008

Time for next Cool Stack release

It's been close to 4 months since we released Cool Stack 1.2, so it's time to start thinking about the next release. Here's what we have planned so far and as always I'm looking for feedback from current and future users on what you'd like to see. Needless to say, that all currently known bugs will be fixed and the current patches will be rolled into the release. However, if you don't tell us about the problems you've run into, we won't be able to fix them. So, once again I'd like to encourage people to please post your problem/issue/tips etc. on the forum.

Here's a list of stuff we're currently looking at for Cool Stack 1.3 :

ComponentVersion in Cool Stack 1.3
Version in Coolstack 1.2
Apache 2.2.8 2.2.6
Tomcat 5.5.26 5.5.23
php 5.2.5 5.2.4
mysql 5.1 5.0.45
squid 3.0 2.6
apc 3.0.16 3.0.14
mod_perl 2.0.3 2.0.2
rails 2.0.2 1.2.3
Add php extensions: memcache , pdflib, and freetype , mcrypt

libevent 1.3e 1.3d
memcached 1.2.5 1.2.2
mod_jk 1.2.26 1.2.25
lighttpd 1.4.18 1.4.16
nginx 0.5.35 -
dtrace ruby extension -
multithreaded perl -
Improve lighttpd with more builtin extensions -
mysql performance improvement with libfasttime/libmtmalloc -
Add ruby gems :postgres, mysql. -
Improve ruby build process (to make easier compilation of external gems) -

I would like feedback on a couple of areas in particular:

  • MySQL 5.1 has many performance improvements and we'd very much like to package this version even though it's not FCS yet (it's been out for a looong time now and should be quite stable). Should we go ahead and use 5.1 instead of 5.0.x ?
  • We've heard some complaints about perl not being multi-threaded, but there is a small performance penalty associated with building it multi-threaded (especially if you're not going to be using it's multi-threaded features).  Any thoughts ?

Of course, please feel free to comment on anything else you'd like seen added/changed as well.




Tuesday Sep 18, 2007

Cool Stack 1.1.1 contents

We are currently working on Cool Stack 1.1.1 which will have the following components updated, in addition to fixing known bugs :

  • Apache 2.2.6 : Add mod_proxy, mod_fcgid, SMF support
  • PHP 5.2.4      : Add dtrace extension, add new fastcgi version for use with non-Apache webservers
  • APC 3.0.14
  • Suhosin 0.9.20
  • MySQL 5.0.45: Add SMF support
  • Perl 5.8.8       : Add Sys:Syslog, DBI, DBD-mysql extensions
  • memcached 1.2.2
  • squid 2.6
  • tomcat 5.5.23
I would love to get feedback on this. Is your favorite module/extension missing ? Are there bugs/issues you have run into that haven't been reported on the forum ?


I'm a Senior Staff Engineer in the Performance & Applications Engineering Group (PAE). This blog focuses on tips to build, configure, tune and measure performance of popular open source web applications on Solaris.


« June 2016