Re: [WDSCI-L] Yet another question about updates

[Mark Phippard <MarkP@xxxxxxxxxxxxxxx>]

Vern,

I am going to punt on the PTF's because I do not know what they are or 
contain.  I suspect it is a way of getting the equivalent of what is in 
that zip file into the IFS of your iSeries, but it is speculation.

I would instead like to explain why the Update process is slow and what 
you can possibly do to make it more bearable.

The lowest level of packaging in Eclipse is the plugin.  There is a 
somewhat artificial layer above plugins called features, which is 
essentially a collection of plugins.  The update manager works with 
features.  When features are installed, they can advertise to Eclipse that 
there is an Update site that it can check for updates to this feature. 
When you do the search for updates, Eclipse contacts each of these sites, 
one at a time, and searches for updates.  The problem, is in what it has 
to do to determine if there are updates.  An update site consists of a 
site.xml file that details all of the features available on that site, and 
then all of the jar files that contain the features and plugins.  The 
search process first downloads this XML file, but it then appears to have 
to download the jar file for each feature specified in the XML file so 
that it can then read the XML file inside that jar file to determine if 
there are any updated plugins that you do not have.  The jar files for the 
features are relatively small, but we are still talking several hundred KB 
that has to be dowloaded.

The jar files for the plugins would not be downloaded unless you actually 
install the updates for that feature, but it appears that it is the 
process of contacting all of these sites and downloading all of these jar 
files that makes it take so long.

What you can do to avoid this process is to download the zipped version of 
the update site from IBM, and then manually initiate the installs via 
Update Manager using what they call a local update site.  Obviously if you 
have a lot of PC's then this makes sense anyway since the updates are only 
downloaded once.

There is usually a correct sequence to installing the updates, however, 
the updates should be properly coded with the correct dependencies that it 
will not allow you to choose the items out of sequence.  The actual 
install process is the same as when you use the automatic method, you are 
just doing it from a local copy.

What I have done in the past is to copy these update sites to a folder 
being served by my internal HTTP server.  This just allows the users to 
setup a permanent local update site that hits the URL of that http server. 
 If you want to do this with IBM's zip file, what you have to do is merge 
all of the jar files in the feature and plugins folders into a single copy 
of those folders, and then likewise merge the contents of all of the 
site.xml files into a single site.xml file.  It is actually easier to do 
than it sounds.

Good luck

Mark