"Don't make me think"...

> [hidden email] wrote:
>
> >
> > 2. Even if you feel tempted to delete the whole "sm" dir, it is
> > sufficient to delete the .sgz files. That way you don't needlessly throw
> > away the package cache. :)
>
> Sometimes the packages themselves were corrupted. Some month ago I found
> some gz or mcz files in the cache containing only some sort of http or
> similar error messages. In that special case the original URLs were
> broken and I had deleted the whole sm directory *before* installing them
> with the SM package loader. The broken packages then produce the "syntax
> error".
>
> Regards,
> Martin

> Let me repeat:
>
> 1. For the above issue - press alt-. and then instead execute
> "SMSqueakMap bootStrap" which will force an upgrade of SM in the  
> image.
> No need to nuke the sm dir.
>
> 2. Even if you feel tempted to delete the whole "sm" dir, it is
> sufficient to delete the .sgz files. That way you don't needlessly  
> throw
> away the package cache. :)

> Hi!
>
> Adrian Lienhard <[hidden email]> wrote:
>> ... is a good usability guideline.
>>
>> Now, this question (see attached screenshot) made me think a lot ;-).
>> And surprising enough, it even proposes "thinking" as one of the
>> answers (as if the developer knew upfront that the user is going to
>> have a hard time to guess right). I have no clue what the right
>> answer should be. Does anybody know what this is all about?
>
> Yes, this is about how an image with the "wrong" classes should  
> load an
> ImageSegment.
> The code asking the questions is not SM itself - it is
> ImageSegment/SmartRefStream code getting into trouble.
>
> Now, the issue is that if you have multiple images sharing the same  
> "sm"
> dir (which is fine in principle) and you update one of those images to
> the current new SM 2.2 - then when the old images try to synch with  
> disk
> (which they do when you fire up the SMLoader) they end up trying to  
> load
> an ImageSegment produced by the SM master server running 3.8 into a  
> 3.7
> image - which is a NO GO as I mentioned earlier here:
>
> http://lists.squeakfoundation.org/pipermail/squeak-dev/2006-April/ 
> 10228
> 6.html
>
> Now, it would be dandy if I had thought about this scenario  
> "yesterday"
> and added code that can handle "new" ImageSegments and do something
> "smart" - unfortunately I didn't have that foresight and I can't
> magically "fix" all images already out there.
>
>> Btw, using latest 3.9a image, and the default SM bugfixing technique
>> (*) solved this problem too.
>
> Let me repeat:
>
> 1. For the above issue - press alt-. and then instead execute
> "SMSqueakMap bootStrap" which will force an upgrade of SM in the  
> image.
> No need to nuke the sm dir.
>
> 2. Even if you feel tempted to delete the whole "sm" dir, it is
> sufficient to delete the .sgz files. That way you don't needlessly  
> throw
> away the package cache. :)
>
>> Cheers
>> Adrian
>
> regards, Göran
>

> On Apr 18, 2006, at 3:37 PM, [hidden email] wrote:
> > Let me repeat:
> >
> > 1. For the above issue - press alt-. and then instead execute
> > "SMSqueakMap bootStrap" which will force an upgrade of SM in the  
> > image.
> > No need to nuke the sm dir.
> >
> > 2. Even if you feel tempted to delete the whole "sm" dir, it is
> > sufficient to delete the .sgz files. That way you don't needlessly  
> > throw
> > away the package cache. :)
>
> Hi Göran,
>
> I'm not sure I follow. Is the above a recipe for getting SM to work  
> in Squeak 3.7?
>
> Thanks,
>
> Colin