Hi Nicolas,

thank you. I think I will also override #default: to ensure that only behaviors are stored. At the moment, the meaning of #default changed without taking take of #default: so that "self default: self default" still works. No updateMap/script should be required if we make sure that ReleaseBuilder also resets all AppRegistries.

Best,
Marcel
Am 06.04.2018 23:12:43 schrieb Nicolas Cellier <nicolas.cellier.aka.nice at gmail.com>:
Back to the original bug, I've fixed it in trunk.


Repairing broken SoundService default requires a new script (done).

Now I don't remember, should I create an update map to make sure that this script is evaluated?


2018-04-06 22:20 GMT+02:00 Nicolas Cellier <nicolas.cellier.aka.nice at gmail.com [mailto:nicolas.cellier.aka.nice at gmail.com]>:

Hi David,
Just a small brook for washing yet another https://en.wikipedia.org/wiki/Labours_of_Hercules#Fifth_labour:_Augean_stables [https://en.wikipedia.org/wiki/Labours_of_Hercules#Fifth_labour:_Augean_stables] ;)


My advice would be to inspect the entries updated these last two years, and if still relevant confirm (or fix) else close.

Fortunately, there are not so much of them.


Then I would close all the rest (resolution = timeout).

Other teams including Pharo have such a policy.


Generally, I hate when my opened issues are closed like that, because I mostly don't use Pharo, so I consider these as pure gifts.
There is a side effect that I'm not encouraged to report anymore, and I rarely reopen them, because I have no time to check.
There is no such thing as gift-support-service.


But, in the end, Pharo team is right: accumulation of rotten issues is not sustainable

- either the issue is important, and it must/will be reopened

- or no one cares and the issue can be closed.


If Heracles himself wants to inspect and resolve the timed-out issues, he would just have to set a filter with a few clicks.

But there's no hurry, the Augean stables had to wait 30 years...


2018-04-06 21:58 GMT+02:00 David T. Lewis <lewis at mail.msen.com [mailto:lewis at mail.msen.com]>:

Nicolas,

It is really good to see those Mantis issues being addressed. Thank you!

Dave


> Hi Marcel, Tobias,
> I encountered a bug, when playing a sound, the default SoundService (class
> inst var) was not a class (BasedSoundService) but an instance (a
> BasedSoundService).
>
> Why?
> Because, there is a TileMorphTest>>setUp that remember which is the
> default, then tries to restore it in tearDown.
> It used to work.
>
> But, in between, SoundService default and defaultOrNil have been
> re-defined
> so as to answer an instance rather than a class.
>
> Thus, when we play the tests, we can no more play the sounds...
>
> I hesitate to open a Mantis issue, no one read nor close them, so the bug
> report is only here for the moment.
>
>





-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20180408/169f1add/attachment.html>