[squeak-dev] Funding (was: What Constitutes a Complete and Final Release?)

2008/4/8 Randal L. Schwartz <[hidden email]>: Count me in :)

As another observation: we should determine somehow the boundaries
which should not be crossed and change dev tools to warn developer
when he crossing these boundaries.
One of most powerful patterns in smalltalk, that we can extend
behavior of base classes like Object
or ProtoObject.
It is good when you just add few new methods, the chance that your
extension will conflict with another package is minimal.
But when you going to override methods or restructure classes(by
adding/removing vars), things getting worser and worser.
Sometimes it is impossible to include a nice workaround or extension
without overriding methods in base(or any other) classes so, ability
to redefine behavior of basic classes should be preserved.
But from the moment when someone needs to change/extend basic things
we getting into trouble of names/versions conflicts when using package
in different images/environment.
This issue, i think, is the main reason why forks appear.
And fork is something which i think is most nasty thing which happens
during development. It leads to fragmentation of developers community,
user base, prevents exchange ideas/improvements and many other things
which i missed :)

If we want Squeak to gain more popularity and flourish, we should make
sure that in future versions a chance of fork will be minimal.
And minimal chance should mean, that developer should have such
image/tools with which he don't needs to make any forks, and be sure
that his application will work rock solid under any circumstances in
any Squeak environment.

--
Best regards,
Igor Stasenko AKA sig.

On Monday 07 April 2008 16:36:29 Sean Heber wrote:

> For what it's worth, as a long time lurker and essentially a non-user
> myself with a background in "traditional" languages, these two
> passages nearly completely nailed how I feel about Squeak and the
> idea of using Squeak as it exists now.  I couldn't have said it
> better myself - and I've tried.  :-)

        I've already been through a lot of this.  What -seems- like it should be
relatively easy is apparently very difficult, like creating an image from
scratch or getting some one to help do it so I can learn.  Nobody seems to
want to take the time.  I'm still interested in using Squeak for Robotics,
but it isn't anywhere near usable for that yet - not enough access to low
level hardware for embedded system use.

        8-Dale
--
I can handle complexity.  It's the simple things that confound me.
The Dynaplex Network - http://www.thedynaplex.org and
Hybrid Robotics - http://www.hybotics.com

--
http://people.squeakfoundation.org/person/mikevdg
http://gulik.pbwiki.com/

Hi--

      Great list, David! Thanks for writing that. I would love to get my
hands on any failing tests you might have, so as to get into specifics.

      Igor writes:

 > Smalltalk, by its nature does not defines a high-level abstractions
 > which can be called 'module' or 'package'.

      I don't think doing so goes against the nature of the system, it
seems to me that the designers just didn't get around to it (another one
for the "historical accidents" category). In particular, I don't think
having an object memory model and having high-level abstractions for
modules are in conflict.

 > I'm aware of at least two of module-based solutions for Squeak:
 > - Spoon by Craig Latta
 > - Namespaces by Michael van Der Gulik , as part of his SecureSqueak
 > project
 >
 > Both systems currently in development. And almost 99% it is solo
 > development by their authors.

      That doesn't surprise me in the case of Spoon, since enabling team
development is the goal. It has to get to a certain usable place first,
although I do think other developers could come in sooner, if we could
work in person until it gets to that place.

 > Why we don't see these systems already employed?

      Well, Spoon isn't ready today. I'm spending Mondays working on a
2008-06-20 release. I'm working on the history system now (an
object-memory-based replacement for the changes and sources files). For
more details, please participate on the Spoon mailing list (also
available via NNTP as [1]), or on IRC in channel #spoon on
irc.freenode.net. I just put out a call for history system use cases on
the mailing list.

 > I think, the main problem is more social than lack of manpower or
 > funding: There is no high pressure from squeak community (and nobody
 > having an ultimate power to force it) to abandon obsolete concepts,
 > sacrifice ST-80 compatibility (partly) and move forward with system
 > based on better design & modularity.

      Personally, as a board member I'd like to lead in that direction
and just give someone else a turn next year if nobody follows. :)


      thanks again,

-C

[1] nntp://news.gmane.org/gmane.comp.lang.smalltalk.squeak.spoon

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

2008/4/8 Michael van der Gulik <[hidden email]>:
Yes, if you read my latest post in this thread about forks. Isn't it
would be better to make fork become Squeak in that way? ;)
I don't want to repeat myself, but do you think it wouldn't be better
to incorporate module-based system in Squeak release?
Because if not, then obviously Squeak community will be forked as
well. Is there a way to put stop on this? Can we make a system which
will make need of forking unnecessary?

> Gulik.
> --
> http://people.squeakfoundation.org/person/mikevdg
> http://gulik.pbwiki.com/
>
>
>



--
Best regards,
Igor Stasenko AKA sig.

Refactoring and bringing new modules online takes a lot of time and
effort. In this present climate such effort is wasted if it is only
directed towards some future squeak, since most squeakers are using
something other than the latest squeak.

Take Rio for example, which has gained some fans, it has been available
for over a year now, and is about ready for an api review. For rio to
become accepted, I would expect ensure that it loads in OLPC, Croquet,
3.7+, I would expect it to take up to 4 years for FileDirectory to expire!

Keith

Gulik.

--
http://people.squeakfoundation.org/person/mikevdg
http://gulik.pbwiki.com/

John M McIntosh wrote:
> Well I'm not a lawyer, however my understanding is below, I'm sure
> others will comment.

Well, it's basically just a question about what are the base packages
covered since I happen to know a couple of them being used by Sophie
that probably aren't BSD licensed (Freetype comes to mind).

> Sophie loads onto a base 3.8.x squeak image (license for that is?)  
> comes from I think the iSqueak repository.
> then we load stuff from
>
> source.impara.de/iSqueak   (license?)  ask impara
> source.impara.de/Tweak   (license?)  ask impara
> source.impara.de/freetype (license?)  ask impara
> source.impara.de/Rome  (license?)  ask impara
> source.impara.de/Grit   (license?) ask impara

Tweak I can actually speak for myself ;-) but for the others it would be
interesting to know what license they're under. I would suspect MIT
because these started during my days at Impara and at that time MIT was
the preferred license. And the core Freetype libs are probably under
whatever license Freetype is under these days (haven't checked in a while).

> I also have stuff from
> http://www.squeaksource.com/XMLSupport.html   (license?) ask Michael
> http://www.squeaksource.com/ToolBuilder.html (license?)

Oh, Toolbuilder. Right. I'll ask the authors for MIT licensing. I think
that for historical reasons it is currently probably under Squeak-L.

> For
> source.sophieproject.org/Sophie  any category starting with Sophie-  is
> clearly written for Sophie thus under the
> Sophie license.  In theory all code in there should be under the Sophie
> license however there *could* be some
> contamination when you consider we modify or add to existing classes
> found in the base squeak and from other repositories on  impara.de

Yeah, I'm more curious about the core classes etc. not as much about any
extensions/overrides.

> Say for example overrides  or mod to the base URI class in Squeak, what
> license does that method have then?

Don't care. Either one is fine as far as I am concerned.

These are actually interesting to me. Any word on XUL?

> Anyone of course relying on this should do their own audit of course.

Yup, but thanks for the overview.

Cheers,
   - Andreas

On Mon, Apr 07, 2008 at 09:38:03PM -0000, kengbrown wrote:
> " http://bugs.squeak.org/view.php?id=6890 "
> Installer mantis ensureFix: '0006890: PluggableListMorph is slow'.
>
> After I install 6890, the System Browser scrollbars no longer work, they extend the full
> length. The scroll arrows appear to work however.

I "fixed" this by removing the fixing of bug 6890 from the
installer script. That fix is a little too invasive yet.

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

> Take Rio for example, which has gained some fans, it has been available
> for over a year now, and is about ready for an api review.

Count me in for that, as I just reimplemented parts of Rio in GNU
Smalltalk (in the sense that I looked at the API document on Squeak and
refactored the code in GST to obey that API).  I only implemented the
basic parts + ZIP access (no recursion yet, no renaming yet), but I do
have some ideas on what's nice and what's less nice.

Paolo

I tried to install Rome in a new image, but failed because of  
dependencies I could not figure out.
I loaded the Rome-* packages one after the other and I also loaded  
FreeType from source.impara.de but I still cannot load Rome-Fonts  
because it depends on the class FreetypeFont. Also, RomeBalloonCanvas  
has references to CIdentityTransform, CMatrixTransform etc. which are  
used for the demos.

Did anybody manage to load Rome in a fresh image? Or is there a loader  
that would do this automatically? (How is the whole Sophie project  
loaded?)

Cheers,
Adrian


On Apr 7, 2008, at 23:47 , John M McIntosh wrote:

>>>>> "Andreas" == Andreas Raab <[hidden email]> writes:

Andreas>  And the core Freetype libs are probably under whatever license
Andreas> Freetype is under these days (haven't checked in a while).

First hit for "freetype license" is
<http://freetype.fis.uniroma2.it/license.html>, which says Freetype is
essentially dual-licensed, BSD-credit and GPL.

--
Randal L. Schwartz - Stonehenge Consulting Services, Inc. - +1 503 777 0095
<[hidden email]> <URL:http://www.stonehenge.com/merlyn/>
Perl/Unix/security consulting, Technical writing, Comedy, etc. etc.
See PerlTraining.Stonehenge.com for onsite and open-enrollment Perl training!

>>>>> "Michael" == Michael van der Gulik <[hidden email]> writes:

Michael> One of the reasons that it is unlikely that my Namespaces
Michael> implementation will be able to be part of the squeak.org image is
Michael> because I plan to do away with the "Smalltalk" SystemDictionary and
Michael> reorganise everything into individual, live Packages, each containing
Michael> hierarchies of Namespaces and Classes. The community here would never
Michael> agree to such a radical change.

I'm jumping in here only partially informed, but I know Cincom has a namespace
mechanism.  Is your framework similar, perhaps even compatible?  And would
there be a reasonable way to file in legacy classes into such a new system?

--
Randal L. Schwartz - Stonehenge Consulting Services, Inc. - +1 503 777 0095
<[hidden email]> <URL:http://www.stonehenge.com/merlyn/>
Perl/Unix/security consulting, Technical writing, Comedy, etc. etc.
See PerlTraining.Stonehenge.com for onsite and open-enrollment Perl training!

On 07.04.2008, at 21:55, Andreas Raab wrote:
The official Rome repository is

        http://squeaksource.com/Rome.html

where it clearly states MIT.

- Bert -

Bert Freudenberg wrote:
> The official Rome repository is
>
>     http://squeaksource.com/Rome.html
>
> where it clearly states MIT.

Excellent, thanks! I had completely forgotten about that repository ;-)

Cheers,
   - Andreas

El 4/8/08 6:15 PM, "Stephen Pair" <[hidden email]> escribió:
Stephen:

Very happy talented Squeaker said same I saying for a while in good English.

Now , who is against this and vote for not start 3.11 ?

Monday meetings with Pavel, Craig, Matthew and a lot more could focus on
license and 4.0, so all could win.

Edgar

Stephen Pair wrote: There are now two mechanisms for doing this, Universes and
Sake/Packages. (see Sake, and Packages on Squeaksource)

The idea of Sake/Packages is that an image includes a class which lists
methods for loading all of the other packages, including their
dependencies. Much the same as Universes, but it is simpler to use and
maintain since it doesnt have any the dependence upon a server, or a UI.

Installer provides command line front end to both.
>  Eventually, maybe we could make it so that certain things can be
> triggered to get loaded automatically (I can imagine for example that
> turning on syntax highlighing could be made to load Shout automatically).
>
> - Stephen
best regards

Keith

-----Original Message-----
From: [hidden email] [mailto:[hidden email]]On Behalf Of Stephen Pair
Sent: 08 April 2008 10:15 PM
To: The general-purpose Squeak developers list
Subject: Re: [squeak-dev] What turns off newcomers

On Mon, Apr 7, 2008 at 8:02 PM, tim Rowledge <[hidden email]> wrote:

I think there is a bit of a logical problem here ins that if the system *didn't* have 'all that junk' the immediate result would be complaints that 'it doesn't have all that useful stuff like other systems'.

I'd prefer this logical problem be solved by moving as much junk as is practically possible out of the base image(s), but make it trivial to find and load them back in.  Half that battle is in the finding.  Right now, I can imagine that it's difficult for many people to tell the most useful stuff that's out there.  Maybe the board or some other body could build a consensus top 20 downloads (for a newbie developer audience at first), give them prominence in the image, and make sure the load ok in the base image(s), that would be a good start.  Eventually, maybe we could make it so that certain things can be triggered to get loaded automatically (I can imagine for example that turning on syntax highlighing could be made to load Shout automatically).

- Stephen

Gary Chambers wrote:
>
> Perhaps a friendly "wizard" upon stating of the fresh image may be
> appropriate. It could use Universes, perhaps, a meta-universe of
> high-level choices ("have fun" "develop" etc.).

"have fun" should be enabled by default, right now it seems to be
disabled ;-)

Michael