Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
[platform-update-dev] Shared Install versus Shared Configuration


Hi Everyone,

I have one more question which I posted to the Equinox newsgroup but got no response.   I know it has been less than a day...but I am trying to get a release together by Friday and I am in a serious time crunch right now so I apologize for all the questions ahead of time.  The post is pasted at the end of this email, but my question, in a nutshell, is as follows:

In the case of Scenario #2 from the Eclipse multi-user install docs (the shared install), how do you go about installing updates to the shared installation area and ensure your users pick them up?  Is it supposed to just work?  I have run across at least one machine where it doesn't, but perhaps I am doing it wrong.  Here is the problem.  From the docs, a shared install scenario implies there is only ever a config.ini in the configuration area contained in the shared install area (in other words it is not initialized).  OK, fine.  But who installs the updates to the plugins when you want to add features to a product?  I think the answer to this question is - it doesn't matter as long as they have write access to the shared installation area.  However, by what mechanism are these changes propagated to the restricted users sharing the installation?  When a restricted user runs Eclipse after the shared install has had more features/plugins installed into it, I assume the changes aren't looked for or detected unless a -clean is issued (although it appears to work strangely enough on all machines I have tried except one).
The logical part of my brain tells me that in order to install updates to a shared install, you should run with your configuration area explicitly set to the configuration area contained in the shared install area, but this will force metadata to be created in this area and by definition you are no longer in a shared install mode anymore.  In fact, Eclipse seems to detect that this metadata is present and automagically flip into cascaded mode (or Scenario #3 - shared configuration) - which is quite different.

Am I missing something here?

Thanks in advance,
Mark.


On Mon, 11 Jun 2007 18:41:10 -0400, Mark Melvin <mark_melvin@xxxxxxxx> wrote:

> Hi There,
>
> I have a problem which is driving me insane.  I'm not sure if this  
> belongs here or if I should be posting this to the platform newsgroup,  
> but here goes.
>
> This is a question about a problem with a product based on a full  
> platform runtime (3.3RC3), but I need help tracing through the OSGI  
> stuff.  I have a weird bug that is only reproducible on one machine (it  
> is not reproducible on many other machines that I have tried).  It  
> involves new features installed into an Eclipse site not being noticed.  
> The features and plugins are there in the features and plugins folders -  
> but Eclipse ignores them.  I am running a "shared install" where the  
> features and plugins were installed by an administrator (using the  
> Eclipse command line update mechanism - called from our installer), but  
> when I log in as a regular user these features are not noticed and my  
> user-specific configuration area goes unmodified.
>
> On every other machine I have tried this on, when I log in as a regular  
> user after performing the "update", my user-specific configuration area  
> is regenerated with the new information and the features are there.  I  
> cannot for the life of me figure out what the heck is going on here.  
> The failing machine is a ghosted test machine that is basically  
> completely clean install of WinXP SP2.  Why it works on every other  
> machine is beyond me.  The only possible difference I can see is that  
> this old machine is FAT-based and all of the others I have tried are  
> NTFS-based.  The only reason I mention this detail is because the act of  
> updating (installing the new feature) seems to create a lot of junk in  
> the Local Settings\Temp folder of the admin account (cached versions of  
> the .JAR files), and the configuration area of the administrator is also  
> modified.  But - this should not matter as far as I understand things.  
> When I fire up Eclipse the next time as a restricted user - shouldn't it  
> do some sort of lightweight scan to determine if the configuration area  
> needs to be refreshed?
>
> Oh, an important note.  If I execute Eclipse with -clean, the feature is  
> detected.  But this should not be necessary in my opinion.  Although now  
> that I think about it...if you do not execute Eclipse with -clean, how  
> does Eclipse *ever* know you have installed additional features when you  
> are running in a "shared install" scenario?  It looks to me like the  
> install information goes into the configuration area of the user that  
> installed the features -  in this case the administrator.  But this  
> doesn't make sense.  How the heck would you *ever* be able to install  
> new features into a shared installation so they were accessible without  
> doing a -clean?  And why the heck does it work for all of my machines  
> except one?  I'm very confused...  ;o)
>
> I guess what I was hoping to find was:
>
> 1) More knowledge on exactly how the configuration areas know when  
> something has been added to the shared install, and thus need to refresh  
> themselves (in particular, regenerate platform.xml).
> 2) A way to further diagnose this problem - perhaps enabling some sort  
> of logging or secret console trick to show me why a feature may not have  
> been loaded.
>
> I mention point #2 because if I do into Manage Configuration, and show  
> disabled features - it shows me the feature that it is failing to find  
> and it says it is disabled.  However, I assume this may be a red herring  
> because it looks like the manage configuration tool scans the features  
> and plugins directories and then tries to match what it found to active  
> bundles.
>
> Anyway, any enlightenment would be most welcome.  I have spent over 2  
> days on this so far.
>
> Thanks,
> Mark.
AMI Semiconductor - "Silicon Solutions for the Real World"
NOTICE: 
This electronic message contains information that may be confidential or privileged. The information is intended for the use of the individual or entity named above. If you are not the intended recipient, please be aware that any disclosure, copying, distribution or use of the contents of this information is prohibited. If you received this electronic message in error, please notify the sender and delete the copy you received.





Back to the top