The "published" package release feature is very nice!

However, I think one more tweak to the package loader would make it 
much more useful.  Currently, you can't easily tell if a package 
release is published or not while browsing with the package loader.  
I'd suggest marking the package releases which are *not* published with 
something like parentheses or italics (in the package list/upper-left 
pane).

So, an expanded package might look like this:

\/ Whisker Browser
    (1.1alpha)
    1.0
    (0.93)

This shows that the 1.1alpha release of Whisker is unpublished, so be 
wary when loading that version.

This is a real-life use case by the way, I wanted to make a new 
unpublished release of Whisker, which I could stuff changes and 
bugfixes into, without worrying about making a new release each time.  
(Such as the bugfix which Russell Penney just posted.)

I think the parentheses (or italics) are a pretty good way of 
indicating that you might not treat the release as a "real" one.  If 
you didn't know what it meant at first, it would become obvious right 
away when you tried to install one and got the "not published" confirm 
dialog.

I was going to ask if this was an easy fix to make, but I just went 
ahead and made it myself. :-)  (Doing parentheses was simpler than 
worrying about whether Text would be properly handled.)  Fix is 
attached.

- Doug


non-compressed snippet is here:

!SMPackageReleaseWrapper methodsFor: 'as yet unclassified' stamp: 'dew 
7/12/2004 16:17'!
asString
	^ item isPublished
		ifTrue: [item smartVersion]
		ifFalse: ['(', item smartVersion, ')']! !



On Sunday, July 11, 2004, at 12:24 PM, goran.krampe at bluefish.se wrote:

> Hi people!
>
> No serious bug reports yet on the new SM Package Loader, perhaps noone
> is using it. :)
>
> Just wanted to mention that the "published" feature should be more
> useful now so maintainers and users need to understand what it IS and
> how to USE it. If you are just a user on SM you can skip down to the
> Short version, maintainers should read this too:
>
> - A release has a checkbox that says "published". If you check that
> checkbox when you create or edit a package release you as a maintainer
> are saying that:
>
> 	1. Ok, this release now works and I will NOT DELETE it in the future.
> 	2. I will also not modify the file that the URL points to, it can be
> considered IMMUTABLE.
>
> ...so until you have checked that checkbox the release can be both
> faulty (it may not even install properly and may hose your image, set
> your house on fire and do other nasty things) and you can delete it
> without notice and also change the file that it references.
>
> This is very convenient for you as a package maintainer - you can make 
> a
> release and test it calmly without being afraid that someone else loads
> it and gets into trouble (unless they are terribly stupid and they can
> then blame themselves), because now the new Package Loader both:
>
> 	- Warns people if they are trying to install a non published release.
> 	- Has a filter to only show published releases.
> 	- Does not use non published releases when doing the "upgrade all
> installed" packages.
>
> Another nice side effect of this is that the cache suddenly can be
> trusted in a whole different way. Published releases can happily be
> cached, because they are immutable.
>
> And yet another thing to know as a maintainer is that when you publish 
> a
> release it is also recorded that YOU did it. This is very good when
> having co-maintainers. :)
>
> Note: There is no enforcement here - as a maintainer you can break
> against the above rules. But it will make people angry with you. :)
>
>
> Short version
> ----------------
>
> Ok, so you have read this far and would like to get the "short 
> version".
>
> If you are a user of SM:
> 	- Don't install releases that aren't published! If you do, you are on
> *your own*.
>
> 	- If you still is foolish or bright enough to dare install or download
> a non published release, remember that it may have changed tomorrow and
> your cache may be invalid - SM doesn't know. So when it eventually *is*
> published you probably want to force download to cache and then install
> it again - which should work with for example Monticello - other 
> formats
> may very well get messed up.
>
> If you are a maintainer:
> 	- You can now put up releases and test them through SM, that they
> install properly and work properly. And when you are sure it works -
> then you can go into SM and check that publish checkbox. This is also a
> nice way to get other people to act as alpha-testers.
>
> 	- Do not publish until you are satisfied with the release because it
> implies that you will NOT delete it in the future, nor will you modify
> the file that it refers to. You may modify the URL though.
>
>
> Now... is that CLEAR? :)
>
> regards, Göran
>
> PS. ...and I just got this report from Reuters that says:
>
> "Over 200 people have signed up with accounts on SM! Yep, 200! Way
> cool."
>
>


-------------- next part --------------
A non-text attachment was scrubbed...
Name: SM2UnpublishedDisplay-dew.1.cs.gz
Type: application/x-gzip
Size: 375 bytes
Desc: not available
Url : http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20040712/85c2dca0/SM2UnpublishedDisplay-dew.1.cs.bin
-------------- next part --------------