[squeak-dev] FractalMorph 1.2 won't load into 3.10.2-7179

Claus Kick wrote: For details - http://installer.pbwiki.com/Squeak311Proposal

The build process produces a "test-candidate" image.

The "test-candidate" is processed into a "release-candidate" image via:

1) All packages are saved to the monticello repository 2) some tools
used in building may be removed 3) version is updated.

The release, the basic image will be fairly functional, elements that
are optional will be easily unloaded. i.e. rather than produce a minimal
image that can be built into a basic image, we will produce a basic
image that can be reduced to a minimal image using Sake/Packages unload
tasks, after which Sake/Packages itself can be unloaded. Rather than
produce an image in which traits are loadable (this would prevent the
core from using traits), we will have an image in which traits are
removable/flattenable.

Derivative images, automatically built from release-basic

   1. -test - basic with tests loaded
   2. -minimal - basic with all removable packages removed, including
LPF, Sake, Installer etc.
   3. -kernel - Pavel's shrink script applied perhaps?
   4. -full/fun - (fun squeak? Edgar?)
   5. -dev - Squeak-dev (Damien is moving over to use Sake/Packages to
build)
   6. -web - Squeak-web
   7. -dev-beta - Squeak-dev beta
   8. -web-beta - Squeak-web beta
   9. -seaside - The seaside one-click-experience
  10. -seaside-magma-pier-magritte-scriptaculous - The basis of my own work
  11. -morphic3 experimental platform

ambitious do you think?

Keith

On Sun, Dec 07, 2008 at 12:18:11PM -0300, Edgar J. De Cleene wrote:
If you don't want to be on the board, then resign and let me. I
think I'm the next candidate by votes

--
Matthew Fulmer -- http://mtfulmer.wordpress.com/

Igor Stasenko wrote:
> One people hate Etoys, others hate Traits. :)
>
> Then lets declare 3.8 a best software ever made and leave things as they is.

How exactly are these two statements related? Why does a dislike for
traits imply that 3.8 is perfect?

> Since 3.8 is perfect, then any contribution made past 3.8 and any will
> be made in future should be rejected because you can't improve what is
> already perfect, right? :)
> Let's then stop discussing how to improve things - because it is pointless.

Is there any point to this rhetoric? If so I fail to see it.

> There's only one thing which makes me uncomfortable: any organism
> which can't evolve and can't react to ever-changing environment
> adequately is doomed to become extinct.

And your point is...? You are the first person to propose stopping to
improve Squeak. Since I know this isn't your goal I can only guess that
you were trying to make some other point which I'm not getting.

Cheers,
   - Andreas

On 07.12.2008, at 16:33, Igor Stasenko wrote:


Traits are actually quite cool. I think people do not hate Traits but  
just object to the way they were added to the system. A similar  
objection applies to how Etoys is intertwined with Morphic, so I'd  
think people don't actually hate Etoys but just that it cannot cleanly  
be removed without severe breakage.

- Bert -

On 7-Dec-2008, at 5:16 PM, Bert Freudenberg wrote:
>
> Traits are actually quite cool.

Traits may be cool, and/or they may just be a fad like I think Aspect  
Oriented Programming is.

However from what I can tell so far Traits are not really Smalltalk-80  
at all.

I Squeak still supposed to be Smalltalk-80?  Is it supposed to be  
Smalltalk + Traits?  Is it supposed to be a standards-compatible  
Smalltalk?  What is Squeak?  Who definitively can answer that these  
days?


> I think people do not hate Traits but just object to the way they  
> were added to the system.

 From what I know at this moment I personally think trying to include  
Traits in the core basic image, and especially any attempt to make use  
of them in the basic core image, is fine just so long as you call the  
result something other than Smalltalk-80.

I do not yet know how Traits have been introduced into Smalltalk, but  
it is my very strong impression that the result is not compatible with  
strict Smalltalk-80.

At this point I (naively) think Smalltalk with Traits _must_ be a fork  
and it _must_ be called something else.  I may be very wrong, and I  
may be starting from the wrong impressions, but that's where my  
understanding takes me to right now.

So, in that sense, I think I would object to the fact they were added  
to the system, not just the way they were added to the system.

I personally would really like Squeak to be a strong, viable,  
Smalltalk-80 implementation with full standards compatibility and with  
a good strong community which provides add-ons, extensions, and such  
as additional packages.  Perhaps for a poor analogy, Squeak should be  
the equivalent of the Linux kernel in GNU/Linux systems, thus allowing  
for variant distributions which might ship ready-to-run images which  
contain specific sets of pre-loaded packages and modifications, but  
which hopefully all derive from the same core image and VM.  A  
slightly better analogy might be the full NetBSD (or FreeBSD) core  
OS.  It's a full base operating system (not just a kernel), but there  
are thousands of additional add-on packages available to any user.  
Even X11 is often considered to be just an add-on package.

--
                                        Greg A. Woods; Planix, Inc.
                                        <[hidden email]>

As far as I am aware Traits are just an implementation detail as far as
Smalltalk-80 compatability is concerned. They may effect the users
ability to understand existing code structure, and provide additional
options for structuring code. Essentially though Traits are transparent
to 99% of client code.

Keith

> I personally would really like Squeak to be a strong, viable, Smalltalk-80
> implementation with full standards compatibility and with a good strong
> community which provides add-ons, extensions, and such as additional
> packages.

Of course, *that* would itself be a fork. Squeak has always been about
experimentation and Smalltalk-80 compatibility was more of a
historical artifact than a design goal.

I'd rather have a great Smalltalk-08 (and soon -09) than something
that matches up with the Blue Book description of Smalltalk-80. The
reason I'm using Pharo now is that it looks like the best near term
'core' Smalltalk.

2008/12/7 Andreas Raab <[hidden email]>:
i refer to Edgar's comment about 3.8 and what is gone 'wrong' after it.

My point :

Is it RIGHT to not do anything , because there is always someone who
will be discontented with it?


> Cheers,
>  - Andreas
>

--
Best regards,
Igor Stasenko AKA sig.

Igor Stasenko wrote:
> 2008/12/7 Andreas Raab <[hidden email]>:
>> How exactly are these two statements related? Why does a dislike for traits
>> imply that 3.8 is perfect?
>
> i refer to Edgar's comment about 3.8 and what is gone 'wrong' after it.

I re-read Edgar's comments again but I still don't see anything in them
that could be construed as framing 3.8 as perfect or denying progress.
Edgar points out (correctly) that 3.8 was the last release that had wide
consensus, he points out (again correctly) that many of the major forks
are 3.8 based. He then goes on saying that 3.9 was "pain for all" and
concludes by pointing out that 3.10 release team was trying to play it
safe. All of it seems to be quite accurate from what I can tell.

> My point :
>
> Is it RIGHT to not do anything , because there is always someone who
> will be discontented with it?

Of course not. That is so obvious it doesn't even bear mentioning. But
then again, has that ever happened? Or is that likely to happen? We have
seen constant improvements in Squeak, mostly non-controversial and in my
experience, the situations where you find great resistance are almost
exclusively those where one side is absolutely unwilling to adopt to
concerns and push things with pseudo justifications like "this is for
your own benefit". If it were, you wouldn't have to force people to use
it - you would make it accessible so that people have the option and
then, when its value is established, you can come back and make a real
case why it should be included by default. This is how the process
should have gone with traits.

Cheers,
   - Andreas

yes.

would it have happened this way, the people implementing traits may have
  noticed that a lot of people did not use them simply because the tools
were not there. IMO the job has been left unfinished, and with the
traits team moving to Pharo we seem to be left with yet another leftover
mostly useless cruft in Squeak, which is the kind of thing these same
people have been creating Pharo to get rid of. how sadly ironic. it
seems to me that a lot of time, energy and good-willingness from all
sides has just been wasted.


Stef

On Dec 6, 2008, at 8:12 AM, Igor Stasenko wrote:

> Hmm, i see nothing fun when i loading random package from squeak map
> have a 50% chance (or less) of successfull load.
> And even if it loads, it could be half-working and may lead to
> DNU/crash each time i using this package.
> Maybe i too dumb , because i can't see how such situation can be
> called wonderfull world for "children's, teachers, researchers and web
> developers".

There was a professor at the University here who had the two Mark  
Guzdial Squeak books on her shelf when I went to work on her computer,  
roughly two years ago.  I struck up a conversation about Squeak.  I  
knew already that she didn't teach Squeak in any courses here.  She  
said Squeak was very neat, but "very buggy."  She found it hard to get  
anything done because she "kept running into bugs."  She sounded as if  
she considered it something of a shame.

- TimJ

Hi Edgar--

 > Hopes of many was when Dan say he wish be on Board.
 >
 > Now I sit on his chair (because maybe nobody with better qualification
 > is at hand ?).

      Dan left the board because he was too busy with work and his other
commitments to participate. You're on the board now because we had two
open slots (Tim had also left), and you were one of the two next highest
vote-getters in the last election.


-C

--
Craig Latta
improvisational musical informaticist
www.netjam.org
Smalltalkers do: [:it | All with: Class, (And love: it)]