As in previous years it was easily worth the trip. Lots of mainstream vendors were present including Sun Microsystems, HP, IBM, RedHat, and Novell. The FreeSoftware Foundation, and the GNOME, Ubuntu, and Fedora projects also had tables. SWAG is back in a big way; buckets of toys and t-shirts all around.
Keynote (Max Spevack)
Good intro that meshed well with the closing key-note. Really a recycling of allot of Open Source philosophy but well presented including recursive humor. You can't have a geek philosophy presentation without recursive humor, GNU does stand for "GNU is not UNIX" after all.
Open Source Legal Issues
(HP)
I was hoping this would explain some specific issues such as what is up with GPLv3 and how one is supposed to handle the intermingling of GPLv2 and GPLv3. But the presentation was more managerially oriented about how to manage the licensing issues and obligations of Open Source. However the presentation was still interesting and comprehensive. HP always does a really nice and thorough job with their presentations, I've never seen one I was disappointed with. IBM and Novell could pick up some good stylistic tips from watching a few HP presentations.
* HP uses OpenLDAP everywhere, specifically Symas.
* Mentions the spoof, sort-of-FUD, that is "indemnification".
- This presentation dove-tailed nicely with closing keynote.
Puppet
(Luke Kanies)
Puppet is an automated configuration management system. Kanies' presentation was well organized and quickly explained the highlights of Puppet, most notably its abstraction layer and how to specify classes and values. I've heard of Puppet before but quickly written it off since it is written in Ruby. If it actually operates and performs as smoothly as this presentation indicates it may be worth it. Ruby still strikes me as a fringe/cult language, and don't we have way too many languages and scripting environments already; it is extremely hard to imagine the world [or my servers] needs yet another one. That and the Ruby-on-Rails crowd are just so off-putting, they remind me of wild-eyed street corner preachers.
* Automatically computes dependencies.
* Uses a single trust domain based on SSL certificate signing.
* Used in the Fedora and CentOS projects.
* Provides an XML-RPC interface (And I love a good API!)
* http://reductivelabs.com/trac/puppet
- My cursory searching so far has indicated that Puppet with also integrate with LDAP (See http://reductivelabs.com/trac/puppet/wiki/LdapNodes)
- SuSE packages from http://download.opensuse.org/repositories/systemmanagement:/misc/
I was hoping for more of an emphasis on the future of OOo, especially some of its architectural deficiencies. But the emphasis was mostly on the best features of the current version; still it was informative.
* The origin of Star was StarWriter for the fondly remembered Z80.
* StarOffice was originally for the Mac.
* The latest release provides a vastly improved, and overdue, improvement to Charts.
* 2.4 offers improvements in notes and ability to strip meta-data.
* 3.0 due around 2008-09 will offer support for ODF 1.2 and M$-Office 12 formats. M$-Office 12 is Microsoft's OOXML. Currently the ODF
version used in OOo doesn't specify the specifics for formulas, thesame deficiency exists in M$'s OOXML specification.
Introduction to Python and/or Solar System Demolition
(Catherine Devlin)
Everyone needs scripting in some capacity and I've found myself using Python for more and more of that; it is just as straight-forward for
parsing and transforming files [mostly CVS] as trusty and venerable AWK and it provides impressive networking/RPC support without having to muck about with the abomination that is CPAN. [I have a strict *NO PERL* policy on all the networks I'm responsible for.] So I figured I might be able to pick up something useful in this one. I however have very low expectations for programming oriented presentations since doing good presentations on programming techniques, languages, environments, etc... is really hard. I've sat through so many interminable and aimless Intro-To-XYZ presentations that I've lost count. But kudos to Ms. Devlin. This presentation was excellent and she is a natural presenter. The little hiccups that *always* occur when presentations include live demonstrations where dealt with deftly and without any break in the linearity of the presentation. That takes talent. The structure of the presentation was nearly perfect and the example script and it's iterations perfect for visually demonstrating the concepts being discussed.
* Using Python's extensive standard library and third party extensions.
* Assessing and defining internal documentation and introspection information.
- I've looked at lots of Python texts and nobody emphasized this before.
* Exception handling was demonstrated; good error handling is something always overlooked in "Intro to" presentations, but it really
should be something addressed right out of the box. And here it was.
* Defining object classes, methods, and accessing superclasses.
- Oddly Python actually has an Objective-C like taste to it regarding superclasses and object methods that I hadn't noticed before.
* Loops as array declarations; very cool!
What's new in PostgreSQL 8.3?
(Robert Treat)
This presentation left me salivating for 8.3. It sounds like 8.3 will provide Open Source a database that can really start to play in the enterprise arena along side Oracle, Informix, and DB2. Lots of good deeply technical content, loved it.
* Why the 8.3 release cycle has dragged on so long.
* HOT (Heap-Only-Tuples)
- http://andyastor.blogspot.com/2007/09/hot-hot-hot.html
- Really raises PostgreSQL up the ladder and should be a big performance win for an already really fast RDBMS.
* Synchronized scans - A concurrent sequential scans of a table will start tag along at the same place as an already running sequential scan of that time and then go back for the part it 'skipped'. Should really help efficiency of tables that are frequently scanned.
* Lazy XIDs - read only transactions don't allocate a transaction id. So the need for VACUUM FULL is greatly reduced and the amount of I/O relating to updating the on-disk meta-data is reduced.
* Multiple auto-vacuum workers.
* Support for UUIDs
* Enum types and declarations.
* Support for GSSAPI
* Integrated full-text-search - no longer a need to futz about with a contrib module.
* Update-able cursors.
* XML support! In database XSLT transforms and XPATH searches.
* Asynchronous commits.
Keynote
(Drew Furtis of FARK, Bradley Kuhn of the Freedom Law Center)
Kuhn provided both an humeros and enlightening summation of various legal issues including the whole SCO vs. IBM thing. It was really
interesting to hear about it from the someone who saw so much of it first hand (and has a solid legal understanding). He had a funny
picture of himself with the guy delivering the summons from SCO. But the talk was also very sobering - the long and short of it was that we
"got lucky" plain-and-simple. SCO found nothing, we weren't really certain there was nothing there to find, and SCO bollixed lots of things
including picking the absolutely worst target imaginable for their initial victim. Hint: don't sue IBM unless you have trainloads of money, and then lots more money to backup that money. Open Source needs less raving about legal issues and more diligence; what the presenter calls the "boring stuff".
After-party
"Dual Core" and NOTACON did a great job. Food was good and lots of free drinks.
My review of OL2007
As in previous years it was easily worth the trip. Lots of mainstream vendors were present including Sun Microsystems, HP, IBM, RedHat, and Novell. The FreeSoftware Foundation, and the GNOME, Ubuntu, and Fedora projects also had tables. SWAG is back in a big way; buckets of toys and t-shirts all around.
Keynote (Max Spevack)
Good intro that meshed well with the closing key-note. Really a recycling of allot of Open Source philosophy but well presented including recursive humor. You can't have a geek philosophy presentation without recursive humor, GNU does stand for "GNU is not UNIX" after all.
Open Source Legal Issues
(HP)
I was hoping this would explain some specific issues such as what is up with GPLv3 and how one is supposed to handle the intermingling of GPLv2 and GPLv3. But the presentation was more managerially oriented about how to manage the licensing issues and obligations of Open Source. However the presentation was still interesting and comprehensive. HP always does a really nice and thorough job with their presentations, I've never seen one I was disappointed with. IBM and Novell could pick up some good stylistic tips from watching a few HP presentations.
* HP uses OpenLDAP everywhere, specifically Symas.
* Mentions the spoof, sort-of-FUD, that is "indemnification".
- This presentation dove-tailed nicely with closing keynote.
Puppet
(Luke Kanies)
Puppet is an automated configuration management system. Kanies' presentation was well organized and quickly explained the highlights of Puppet, most notably its abstraction layer and how to specify classes and values. I've heard of Puppet before but quickly written it off since it is written in Ruby. If it actually operates and performs as smoothly as this presentation indicates it may be worth it. Ruby still strikes me as a fringe/cult language, and don't we have way too many languages and scripting environments already; it is extremely hard to imagine the world [or my servers] needs yet another one. That and the Ruby-on-Rails crowd are just so off-putting, they remind me of wild-eyed street corner preachers.
* Automatically computes dependencies.
* Uses a single trust domain based on SSL certificate signing.
* Used in the Fedora and CentOS projects.
* Provides an XML-RPC interface (And I love a good API!)
* http://reductivelabs.com/trac/puppet
- My cursory searching so far has indicated that Puppet with also integrate with LDAP (See http://reductivelabs.com/trac/puppet/wiki/LdapNodes)
- SuSE packages from http://download.opensuse.org/repositories/systemmanagement:/misc/
OpenOffice.org: Past, Present, & Future
(Andrew Pitonyak)
I was hoping for more of an emphasis on the future of OOo, especially some of its architectural deficiencies. But the emphasis was mostly on the best features of the current version; still it was informative.
* The origin of Star was StarWriter for the fondly remembered Z80.
* StarOffice was originally for the Mac.
* The latest release provides a vastly improved, and overdue, improvement to Charts.
* 2.4 offers improvements in notes and ability to strip meta-data.
* 3.0 due around 2008-09 will offer support for ODF 1.2 and M$-Office 12 formats. M$-Office 12 is Microsoft's OOXML. Currently the ODF
version used in OOo doesn't specify the specifics for formulas, thesame deficiency exists in M$'s OOXML specification.
Introduction to Python and/or Solar System Demolition
(Catherine Devlin)
Everyone needs scripting in some capacity and I've found myself using Python for more and more of that; it is just as straight-forward for
parsing and transforming files [mostly CVS] as trusty and venerable AWK and it provides impressive networking/RPC support without having to muck about with the abomination that is CPAN. [I have a strict *NO PERL* policy on all the networks I'm responsible for.] So I figured I might be able to pick up something useful in this one. I however have very low expectations for programming oriented presentations since doing good presentations on programming techniques, languages, environments, etc... is really hard. I've sat through so many interminable and aimless Intro-To-XYZ presentations that I've lost count. But kudos to Ms. Devlin. This presentation was excellent and she is a natural presenter. The little hiccups that *always* occur when presentations include live demonstrations where dealt with deftly and without any break in the linearity of the presentation. That takes talent. The structure of the presentation was nearly perfect and the example script and it's iterations perfect for visually demonstrating the concepts being discussed.
* Using Python's extensive standard library and third party extensions.
* Assessing and defining internal documentation and introspection information.
- I've looked at lots of Python texts and nobody emphasized this before.
* Exception handling was demonstrated; good error handling is something always overlooked in "Intro to" presentations, but it really
should be something addressed right out of the box. And here it was.
* Defining object classes, methods, and accessing superclasses.
- Oddly Python actually has an Objective-C like taste to it regarding superclasses and object methods that I hadn't noticed before.
* Loops as array declarations; very cool!
What's new in PostgreSQL 8.3?
(Robert Treat)
This presentation left me salivating for 8.3. It sounds like 8.3 will provide Open Source a database that can really start to play in the enterprise arena along side Oracle, Informix, and DB2. Lots of good deeply technical content, loved it.
* Why the 8.3 release cycle has dragged on so long.
* HOT (Heap-Only-Tuples)
- http://andyastor.blogspot.com/2007/09/hot-hot-hot.html
- Really raises PostgreSQL up the ladder and should be a big performance win for an already really fast RDBMS.
* Synchronized scans - A concurrent sequential scans of a table will start tag along at the same place as an already running sequential scan of that time and then go back for the part it 'skipped'. Should really help efficiency of tables that are frequently scanned.
* Lazy XIDs - read only transactions don't allocate a transaction id. So the need for VACUUM FULL is greatly reduced and the amount of I/O relating to updating the on-disk meta-data is reduced.
* Multiple auto-vacuum workers.
* Support for UUIDs
* Enum types and declarations.
* Support for GSSAPI
* Integrated full-text-search - no longer a need to futz about with a contrib module.
* Update-able cursors.
* XML support! In database XSLT transforms and XPATH searches.
* Asynchronous commits.
Keynote
(Drew Furtis of FARK, Bradley Kuhn of the Freedom Law Center)
Kuhn provided both an humeros and enlightening summation of various legal issues including the whole SCO vs. IBM thing. It was really
interesting to hear about it from the someone who saw so much of it first hand (and has a solid legal understanding). He had a funny
picture of himself with the guy delivering the summons from SCO. But the talk was also very sobering - the long and short of it was that we
"got lucky" plain-and-simple. SCO found nothing, we weren't really certain there was nothing there to find, and SCO bollixed lots of things
including picking the absolutely worst target imaginable for their initial victim. Hint: don't sue IBM unless you have trainloads of money, and then lots more money to backup that money. Open Source needs less raving about legal issues and more diligence; what the presenter calls the "boring stuff".
After-party
"Dual Core" and NOTACON did a great job. Food was good and lots of free drinks.