Final 3.6 Plan

Doug Way dway at riskmetrics.com
Fri Jun 6 20:07:46 UTC 2003


Jecel Assumpcao Jr wrote:

>On Friday 06 June 2003 14:14, Tim Rowledge wrote:
>  
>
>>I claim that the full image should be produced by loading packages
>>into the Basic image. What would be in an update stream for Full?
>>Would it even be sensible to consider?
>>
>>I'm going to posit that the audience for premade Full images is
>>people that want to 'play' and use tools like eToys. Perhaps we
>>should consider whether such an audience would be interested in
>>updating - maybe just grabbing the latest stable Full image would be
>>better for them?
>>

Yes.  We talked about this a bit awhile ago, and came to the conclusion 
that maintaining parallel update streams would be too much of a 
maintenance headache.

>Perhaps if a Full Image user tried to update he could get all the 
>changes for the Basic image and then, after checking with SM, any 
>packages that have newer versions would be unloaded from the image, 
>downloaded from the net and reloaded into the image. That should 
>typically be much faster than downloading a whole new image.
>  
>

It is true that, right now, if someone wanted to have a Full image 
follow the update stream, they could take the current 3.6a-5247 image, 
load the "Upgrade to 3.6a Full" script on SM to get the Full image, and 
then simply continue loading updates from the server.  This would mostly 
work, except the re-loaded packages would not be updated.  They would 
need to reload the Upgrade script at that point. (assuming all of the 
external packages are kept up-to-the-minute current, which will probably 
often not be true)

(Unloading packages is not really supported right now, btw, so that step 
can be skipped. ;-)  But generally loading a new version of a package 
over an old one is expected to work.)

- Doug Way




More information about the Squeak-dev mailing list