[V3dot10] Re: Re: [Bug] 0006839: #comeFullyUpOnReload: not sent to MorphExtension's otherProperties (anIdentityDictionary)

[Klaus D. Witzel]

On Thu, 10 Jan 2008 11:34:06 +0100, Edgar J. De Cleene wrote:

> El 1/10/08 5:25 AM, "Klaus D. Witzel" escribió:
>
>> Then the #==2#= change can happen again (at any time and to/by
>> whomsoever). Would you please point me to the change(s) involved so  
>> that a
>> test can be produced which goes red if the wrong #==2#= is changed  
>> again.
>>
>> TIA
>>
>> /Klaus
>
> Complete history about this.
> http://bugs.squeak.org/view.php?id=2788 is the original bug report, you
> could see here the Dan propose and packages build for 3.10 at the time I
> take it and do the change.

Thank you.

> Jerome (wiz) soon discover is wrong change ALL #== to #=
> Also you could see Ralph analysis about which test fail.
>
> We discuss here how to deal this and decide go back.
> All .cs and .mcz was erased and news .cs and .mcz was produced.
> Also a new image.
> The short history is Squeak3.10.gamma.7159.image don't have it and the  
> 172
> changes should be break in several ones (not in only one as I do).
>
> No derived image (web, fun, others) should use any different to
> Squeak3.10.gamma.7159.image.
>
> This is now our base image, should be declared final, as was in public  
> for a
> month ( 06-Dec-2007 20:59  is date on server)
>
> The #== to #= don't happen again if people learn the hard lesson.

I cannot agree with "don't happen again". a) someone declares #== <small  
constant> a code smell; b) a tool highlights the smell while someone else  
is about to change *whatsoever*; c) unhappy days are here again (not sung  
by the Comedian Harmonists).

/Klaus

> No bug report which give us more of 3 touched packages , one of this a  
> Test
> for one of the others two, should go to updates.
>
> Read the "history.morph" from earlier Squeak 3.10 and the two troubles  
> was
> in this situation.
>
> I hope the meet with Colin in Buenos Aires gives us a better MC2 and next
> Squeak could be build with MC2.
> Or with DeltaStreams
>
> Never more with Monticello, two releases by two different teams show is  
> not
> the tool for the job.
>
> Edgar