On 10 March 2010 23:20, Nicolas Cellier
<nicolas.cellier.aka.nice at gmail.com> wrote:
> 2010/3/10 Torsten Bergmann <astares at gmx.de>:
>> Nicolas Cellier wrote:
>>>Then I want you to play a little game:
>>>take the exact Installer script that installed my mantis patches over
>>>3.10, apply it to latest trunk, Cuis and Pharo, and report how it
>>>worked.
>>
>> Nicolas,
>>
>> forget to play such games. I already requested Keith to play such a game
>> and demonstrate his technology to me (see bottom of [1]). The answer was
>> totally unsatisfying (see bottom of [2]).
>>
>> IMHO it will never succeed as long as the original author is not
>> able to demonstrate or even explain in a way so we could really follow.
>> Maybe Ken can ...
>>
>> Bye
>> T.
>>
>
> Be sure, I won't play myself, I know very well the result.
> That's just to stop false assertions to be repeated over and over again.
> The patches have a limited lifetime. Code lives and dies.
> It dies because it has good reasons to.
> Unless you think trunk evolutions were all gratuitous and we should
> better revert ?
>
> One easy demonstration would be to traverse trunk MC repository and
> count how many times each method were changed...
>
> For external packages, i want to believe the Installer scripts works
> well. But not for trunk...
> It could work in a continuous integration scheme but then it's not
> really far from trunk process, just with less convenient tools for
> multi-developper.
> And don't tell cross fork patches come for free.
>
> Keith is right when he says trunk dev. model is not easy to duplicate.
> He is right because changes span over several packages, and gathering
> information is tedious.
> Maybe Pharo SLICE policy improve the model a bit :
> - publish a bug report
> - make a fix in Pharo
> - create a SLICE package, let it depend on all dirty packages, publish
> the SLICE.
> If harvesters wait too long, then the SLICE is obsolete and can
> conflict with other changes.
> MC merge feature helps in this case. Which Installer tool does ?
>
> I'm not a MC-fanatic, it has some weaknesses.
> The worst thing regarding MC is whenever class and methods move into
> another package, so it certainly must be improved.
> But I don't buy changeSets, I already know the limits too well.
>
... which brings us back to a DeltaStreams idea :)

> Nicolas
>
>> [1] http://lists.squeakfoundation.org/pipermail/vm-dev/2010-February/003871.html
>> [1] http://lists.squeakfoundation.org/pipermail/vm-dev/2010-February/003875.html
>> --
>> GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT!
>> Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01
>>
>>
>
>



-- 
Best regards,
Igor Stasenko AKA sig.