Xtreams , Ometa and related

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
28 messages Options
12
Reply | Threaded
Open this post in threaded view
|

Xtreams , Ometa and related

Edgar De Cleene
People:

I working in my SqueakRosCore fork and think if Xtreams should be on it or
not.

Google surfing brings me

http://comments.gmane.org/gmane.comp.lang.smalltalk.squeak.general/156355

So, four years later I wish feedback about this.

Cool working examples if you know of any.

Opinions:

SqueakSource have two projects about Xtreams, which one I should have ?

Ometa  pay same day the work for having it into the .image ?


Edgar
@morplenauta en twitter





_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Phil B
Edgar,


On Sat, 2015-05-16 at 05:41 -0300, Edgar J. De Cleene wrote:

> People:
>
> I working in my SqueakRosCore fork and think if Xtreams should be on it or
> not.
>
> Google surfing brings me
>
> http://comments.gmane.org/gmane.comp.lang.smalltalk.squeak.general/156355
>
> So, four years later I wish feedback about this.
>
> Cool working examples if you know of any.
>
> Opinions:
>
> SqueakSource have two projects about Xtreams, which one I should have ?
>

Unless someone on squeak-dev says otherwise, I would probably go with
the one pointed to by the Xtreams project page
(https://code.google.com/p/xtreams/):
http://www.squeaksource.com/@1LaiqtNjXeSo_VvX/Wfk5wECe

> Ometa  pay same day the work for having it into the .image ?
>

Ometa is a very cool package though I found it a bit temperamental to
get and keep running.  Keep in mind that you need to patch the compiler
which will be an ongoing maintenance issue.

While both packages are very nice, I probably wouldn't want them in a
core image as I tend to prefer the minimalist approach to images.  They
would be great optional packages to have though.  (I understand you're
talking about SqueakROS and not Cuis, but just wanted to offer my
opinion since you were asking what people thought.)

>
> Edgar
> @morplenauta en twitter

Thanks,
Phil


_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Edgar De Cleene



On 5/16/15, 7:24 AM, "Phil (list)" <[hidden email]> wrote:

> While both packages are very nice, I probably wouldn't want them in a
> core image as I tend to prefer the minimalist approach to images.  They
> would be great optional packages to have though.  (I understand you're
> talking about SqueakROS and not Cuis, but just wanted to offer my
> opinion since you were asking what people thought.)

Thanks for your valuable feedback.

Minimalist approach is one of my goals and was here for learn of Juan and
all people here.

So, not in the image but easily load and work.

I asking for Xtreams as have a chance to meet Martin Kobetic when he was
here in Argentina and presented his work.

Seems wonderful at the time and revisiting
https://code.google.com/p/xtreams/ still is.

Could have sense a Cuis version ?

Edgar



_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Phil B
On Sat, 2015-05-16 at 09:59 -0300, Edgar J. De Cleene wrote:

>
>
> On 5/16/15, 7:24 AM, "Phil (list)" <[hidden email]> wrote:
>
> > While both packages are very nice, I probably wouldn't want them in a
> > core image as I tend to prefer the minimalist approach to images.  They
> > would be great optional packages to have though.  (I understand you're
> > talking about SqueakROS and not Cuis, but just wanted to offer my
> > opinion since you were asking what people thought.)
>
> Thanks for your valuable feedback.
>
> Minimalist approach is one of my goals and was here for learn of Juan and
> all people here.
>
> So, not in the image but easily load and work.
>

That would be my suggestion.
 
> I asking for Xtreams as have a chance to meet Martin Kobetic when he was
> here in Argentina and presented his work.
>
> Seems wonderful at the time and revisiting
> https://code.google.com/p/xtreams/ still is.
>
> Could have sense a Cuis version ?
>

I keep telling myself I'll get around to getting Ometa working in Cuis
again but never seem to find the time...

> Edgar
>
>
>
> _______________________________________________
> Cuis mailing list
> [hidden email]
> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org



_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Juan Vuletich-4
In reply to this post by Edgar De Cleene
On 5/16/2015 9:59 AM, Edgar J. De Cleene wrote:

>
>
> On 5/16/15, 7:24 AM, "Phil (list)"<[hidden email]>  wrote:
>
>> While both packages are very nice, I probably wouldn't want them in a
>> core image as I tend to prefer the minimalist approach to images.  They
>> would be great optional packages to have though.  (I understand you're
>> talking about SqueakROS and not Cuis, but just wanted to offer my
>> opinion since you were asking what people thought.)
> Thanks for your valuable feedback.
>
> Minimalist approach is one of my goals and was here for learn of Juan and
> all people here.
>
> So, not in the image but easily load and work.
>
> I asking for Xtreams as have a chance to meet Martin Kobetic when he was
> here in Argentina and presented his work.
>
> Seems wonderful at the time and revisiting
> https://code.google.com/p/xtreams/ still is.
>
> Could have sense a Cuis version ?
>
> Edgar

Yes, Xtreams would be nice to have.

I have no opinion on Ometa, as I never digged into it.

Cheers,
Juan Vuletich

_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Casey Ransberger-2
In reply to this post by Edgar De Cleene
Hey man! Nice to see you on the list again.

I have no particular interest in Xstreams or whatever, though I have noted the popularity of the thing. Making popular things work in Cuis will be good for Cuis.

However: if you get OMeta working cleanly in Cuis, I will personally wash your feet with my mohawk. I mean someone else will have to cover the air fair, but I will -- shit you not -- wash your feet with my freaking mohawk.

There isn't a single thing I'm more interested in.

--C

> On May 16, 2015, at 1:41 AM, "Edgar J. De Cleene" <[hidden email]> wrote:
>
> People:
>
> I working in my SqueakRosCore fork and think if Xtreams should be on it or
> not.
>
> Google surfing brings me
>
> http://comments.gmane.org/gmane.comp.lang.smalltalk.squeak.general/156355
>
> So, four years later I wish feedback about this.
>
> Cool working examples if you know of any.
>
> Opinions:
>
> SqueakSource have two projects about Xtreams, which one I should have ?
>
> Ometa  pay same day the work for having it into the .image ?
>
>
> Edgar
> @morplenauta en twitter
>
>
>
>
>
> _______________________________________________
> Cuis mailing list
> [hidden email]
> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org

_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Edgar De Cleene



On 5/16/15, 5:40 PM, "Casey Ransberger" <[hidden email]> wrote:

> However: if you get OMeta working cleanly in Cuis, I will personally wash your
> feet with my mohawk. I mean someone else will have to cover the air fair, but
> I will -- shit you not -- wash your feet with my freaking mohawk.
>
> There isn't a single thing I'm more interested in.

Hello again Casey.

So you say OMeta could pay the work...

Could ellaborate some use cases?

If I could put OMeta in Cuis hope you could come to Argentina to some future
FAST Smalltalks event.

And we exchange a couple of beers :=)



Edgar
@morplenauta en twitter




_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Casey Ransberger-2
It's a domain specific language for doing various parts of compiler design. If you wanted to experiment with a language idea, it's convenient. It also let's you do things like borrow rules from other grammars, allows your grammar to inherit its behavior from other grammars.

You can play with the Javascript version here...

http://www.tinlizzie.org/ometa-js/#Sample_Project

> On May 17, 2015, at 2:59 AM, "Edgar J. De Cleene" <[hidden email]> wrote:
>
>
>
>
>> On 5/16/15, 5:40 PM, "Casey Ransberger" <[hidden email]> wrote:
>>
>> However: if you get OMeta working cleanly in Cuis, I will personally wash your
>> feet with my mohawk. I mean someone else will have to cover the air fair, but
>> I will -- shit you not -- wash your feet with my freaking mohawk.
>>
>> There isn't a single thing I'm more interested in.
>
> Hello again Casey.
>
> So you say OMeta could pay the work...
>
> Could ellaborate some use cases?
>
> If I could put OMeta in Cuis hope you could come to Argentina to some future
> FAST Smalltalks event.
>
> And we exchange a couple of beers :=)
>
>
>
> Edgar
> @morplenauta en twitter
>
>
>
>
> _______________________________________________
> Cuis mailing list
> [hidden email]
> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org

_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Phil B
In reply to this post by Edgar De Cleene
On Sat, 2015-05-16 at 05:41 -0300, Edgar J. De Cleene wrote:

> People:
>
> I working in my SqueakRosCore fork and think if Xtreams should be on it or
> not.
>
> Google surfing brings me
>
> http://comments.gmane.org/gmane.comp.lang.smalltalk.squeak.general/156355
>
> So, four years later I wish feedback about this.
>
> Cool working examples if you know of any.
>
> Opinions:
>
> SqueakSource have two projects about Xtreams, which one I should have ?
>
> Ometa  pay same day the work for having it into the .image ?
>

OK, it's all coming back to me as to why my port of OMeta2 died:  I had
it working perfectly in Cuis 1.0.  I think I also had it working in Cuis
2.0 (I even have a .pck which I *believe* was kinda, sorta had working
at one point... but now it can only be read in, it no longer works).

IIRC, the issue was that part of OMeta is written in OMeta so you can't
just load it in the way you would a regular package.  I think original
work by Alessandro was a .st file which had a very specific ordering of
code to load and ran some code as part of the filein which installed
OMeta into the image.  The SqueakSource package addressed the loading
issues by splitting it in 2: a preload and postload package.  I believe
the preload package patches the parser so that the postload package can
even be parsed by Squeak.  Then there is some code that gets called
after the postload package has been read in that actually patches OMeta
into the Squeak compiler.  It's a bit tricky and changes to the
Smalltalk parser/compiler since then will require some changes (I
believe it broke as a result of the changes for closure support being
added ~2012 or so)

I'm not sure how far you've gotten with it on your own already... if
what I have might be helpful, let me know and I'll see if I can't clean
it up a bit and put it up on github.

>
> Edgar
> @morplenauta en twitter

Thanks,
Phil


_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Edgar De Cleene
Your info is very useful.
The experiment until now is I can load OMeta in Squeak 4.6 from SqueakSource.
In my fork wich is Cuis 3,2 can’t load and several tricks for having the classes ordered before OMeta uses each failed.
So if you have the original .st of Alessandro , wish have it for try.
In Cuis as cloned from Github the preload package loads fine and the examples works as in Squeak 4.6.
I send mail here with link to .jpg picture.
But post load fails.

Edgar

-- 
Edgar J. De Cleene
Sent with Sparrow

On Tuesday, May 19, 2015 at 6:55 PM, Phil (list) wrote:

On Sat, 2015-05-16 at 05:41 -0300, Edgar J. De Cleene wrote:
People:

I working in my SqueakRosCore fork and think if Xtreams should be on it or
not.

Google surfing brings me


So, four years later I wish feedback about this.

Cool working examples if you know of any.

Opinions:

SqueakSource have two projects about Xtreams, which one I should have ?

Ometa pay same day the work for having it into the .image ?

OK, it's all coming back to me as to why my port of OMeta2 died: I had
it working perfectly in Cuis 1.0. I think I also had it working in Cuis
2.0 (I even have a .pck which I *believe* was kinda, sorta had working
at one point... but now it can only be read in, it no longer works).

IIRC, the issue was that part of OMeta is written in OMeta so you can't
just load it in the way you would a regular package. I think original
work by Alessandro was a .st file which had a very specific ordering of
code to load and ran some code as part of the filein which installed
OMeta into the image. The SqueakSource package addressed the loading
issues by splitting it in 2: a preload and postload package. I believe
the preload package patches the parser so that the postload package can
even be parsed by Squeak. Then there is some code that gets called
after the postload package has been read in that actually patches OMeta
into the Squeak compiler. It's a bit tricky and changes to the
Smalltalk parser/compiler since then will require some changes (I
believe it broke as a result of the changes for closure support being
added ~2012 or so)

I'm not sure how far you've gotten with it on your own already... if
what I have might be helpful, let me know and I'll see if I can't clean
it up a bit and put it up on github.


Edgar
@morplenauta en twitter

Thanks,
Phil


_______________________________________________
Cuis mailing list


_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Phil B
On Tue, 2015-05-19 at 19:12 -0300, Edgar J. De Cleene wrote:

> Your info is very useful.
> The experiment until now is I can load OMeta in Squeak 4.6 from
> SqueakSource.
> In my fork wich is Cuis 3,2 can’t load and several tricks for having
> the classes ordered before OMeta uses each failed.
> So if you have the original .st of Alessandro , wish have it for try.
> In Cuis as cloned from Github the preload package loads fine and the
> examples works as in Squeak 4.6.
> I send mail here with link to .jpg picture.
> But post load fails.
>

OK, I've almost got it working again except for one tiny thing (that
breaks everything)... #compilerClass no longer exists.  I've poked
around and all of the compiler indirection has been removed (i.e.
Compiler is hard-coded everywhere).  This completely breaks OMeta2 so
unless Juan would consider bringing this functionality back, it's
probably not going to work without essentially a rewrite.

... so that's why I gave up on it :-(

>
> Edgar
>
>
> --
> Edgar J. De Cleene
> Sent with Sparrow
>
>
> On Tuesday, May 19, 2015 at 6:55 PM, Phil (list) wrote:
>
> > On Sat, 2015-05-16 at 05:41 -0300, Edgar J. De Cleene wrote:
> > > People:
> > >
> > >
> > > I working in my SqueakRosCore fork and think if Xtreams should be
> > > on it or
> > > not.
> > >
> > >
> > > Google surfing brings me
> > >
> > >
> > > http://comments.gmane.org/gmane.comp.lang.smalltalk.squeak.general/156355
> > >
> > >
> > > So, four years later I wish feedback about this.
> > >
> > >
> > > Cool working examples if you know of any.
> > >
> > >
> > > Opinions:
> > >
> > >
> > > SqueakSource have two projects about Xtreams, which one I should
> > > have ?
> > >
> > >
> > > Ometa pay same day the work for having it into the .image ?
> >
> >
> > OK, it's all coming back to me as to why my port of OMeta2 died: I
> > had
> > it working perfectly in Cuis 1.0. I think I also had it working in
> > Cuis
> > 2.0 (I even have a .pck which I *believe* was kinda, sorta had
> > working
> > at one point... but now it can only be read in, it no longer works).
> >
> >
> > IIRC, the issue was that part of OMeta is written in OMeta so you
> > can't
> > just load it in the way you would a regular package. I think
> > original
> > work by Alessandro was a .st file which had a very specific ordering
> > of
> > code to load and ran some code as part of the filein which installed
> > OMeta into the image. The SqueakSource package addressed the loading
> > issues by splitting it in 2: a preload and postload package. I
> > believe
> > the preload package patches the parser so that the postload package
> > can
> > even be parsed by Squeak. Then there is some code that gets called
> > after the postload package has been read in that actually patches
> > OMeta
> > into the Squeak compiler. It's a bit tricky and changes to the
> > Smalltalk parser/compiler since then will require some changes (I
> > believe it broke as a result of the changes for closure support
> > being
> > added ~2012 or so)
> >
> >
> > I'm not sure how far you've gotten with it on your own already... if
> > what I have might be helpful, let me know and I'll see if I can't
> > clean
> > it up a bit and put it up on github.
> >
> >
> > >
> > >
> > > Edgar
> > > @morplenauta en twitter
> >
> >
> > Thanks,
> > Phil
> >
> >
> >
> >
> > _______________________________________________
> > Cuis mailing list
> > [hidden email]
> > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
>
>
> _______________________________________________
> Cuis mailing list
> [hidden email]
> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org



_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Hannes Hirzel
Phil

Thank you for the report where you get stuck when porting OMeta to Cuis.

#compilerClass

Would it be possible to re-add #compilerClass through a compatibility package?

https://github.com/Cuis-Smalltalk/Cuis-Smalltalk-Dev/tree/master/CompatibilityPackages

--Hannes



On 5/20/15, Phil (list) <[hidden email]> wrote:

> On Tue, 2015-05-19 at 19:12 -0300, Edgar J. De Cleene wrote:
>> Your info is very useful.
>> The experiment until now is I can load OMeta in Squeak 4.6 from
>> SqueakSource.
>> In my fork wich is Cuis 3,2 can’t load and several tricks for having
>> the classes ordered before OMeta uses each failed.
>> So if you have the original .st of Alessandro , wish have it for try.
>> In Cuis as cloned from Github the preload package loads fine and the
>> examples works as in Squeak 4.6.
>> I send mail here with link to .jpg picture.
>> But post load fails.
>>
>
> OK, I've almost got it working again except for one tiny thing (that
> breaks everything)... #compilerClass no longer exists.  I've poked
> around and all of the compiler indirection has been removed (i.e.
> Compiler is hard-coded everywhere).  This completely breaks OMeta2 so
> unless Juan would consider bringing this functionality back, it's
> probably not going to work without essentially a rewrite.
>
> ... so that's why I gave up on it :-(
>
>>
>> Edgar
>>
>>
>> --
>> Edgar J. De Cleene
>> Sent with Sparrow
>>
>>
>> On Tuesday, May 19, 2015 at 6:55 PM, Phil (list) wrote:
>>
>> > On Sat, 2015-05-16 at 05:41 -0300, Edgar J. De Cleene wrote:
>> > > People:
>> > >
>> > >
>> > > I working in my SqueakRosCore fork and think if Xtreams should be
>> > > on it or
>> > > not.
>> > >
>> > >
>> > > Google surfing brings me
>> > >
>> > >
>> > > http://comments.gmane.org/gmane.comp.lang.smalltalk.squeak.general/156355
>> > >
>> > >
>> > > So, four years later I wish feedback about this.
>> > >
>> > >
>> > > Cool working examples if you know of any.
>> > >
>> > >
>> > > Opinions:
>> > >
>> > >
>> > > SqueakSource have two projects about Xtreams, which one I should
>> > > have ?
>> > >
>> > >
>> > > Ometa pay same day the work for having it into the .image ?
>> >
>> >
>> > OK, it's all coming back to me as to why my port of OMeta2 died: I
>> > had
>> > it working perfectly in Cuis 1.0. I think I also had it working in
>> > Cuis
>> > 2.0 (I even have a .pck which I *believe* was kinda, sorta had
>> > working
>> > at one point... but now it can only be read in, it no longer works).
>> >
>> >
>> > IIRC, the issue was that part of OMeta is written in OMeta so you
>> > can't
>> > just load it in the way you would a regular package. I think
>> > original
>> > work by Alessandro was a .st file which had a very specific ordering
>> > of
>> > code to load and ran some code as part of the filein which installed
>> > OMeta into the image. The SqueakSource package addressed the loading
>> > issues by splitting it in 2: a preload and postload package. I
>> > believe
>> > the preload package patches the parser so that the postload package
>> > can
>> > even be parsed by Squeak. Then there is some code that gets called
>> > after the postload package has been read in that actually patches
>> > OMeta
>> > into the Squeak compiler. It's a bit tricky and changes to the
>> > Smalltalk parser/compiler since then will require some changes (I
>> > believe it broke as a result of the changes for closure support
>> > being
>> > added ~2012 or so)
>> >
>> >
>> > I'm not sure how far you've gotten with it on your own already... if
>> > what I have might be helpful, let me know and I'll see if I can't
>> > clean
>> > it up a bit and put it up on github.
>> >
>> >
>> > >
>> > >
>> > > Edgar
>> > > @morplenauta en twitter
>> >
>> >
>> > Thanks,
>> > Phil
>> >
>> >
>> >
>> >
>> > _______________________________________________
>> > Cuis mailing list
>> > [hidden email]
>> > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
>>
>>
>> _______________________________________________
>> Cuis mailing list
>> [hidden email]
>> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
>
>
>
> _______________________________________________
> Cuis mailing list
> [hidden email]
> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
>

_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Casey Ransberger-2
In reply to this post by Phil B
Hi Phil.

Any particular reason compiler indirection can't go into one of our Squeak compat packs?

Got myself injured, and will be a week or three before I can sit at a computer for an extended period of time, but I could have a look at it if there isn't some reason we can't get it into compat. I surfed all the way through Cuis looking at every single reference to Color once. Seems like doing the same wouldn't be so hard looking for refs to Compiler and fixing them as I went.

I guess the question is, can a pck modify core classes?

--C

P.S.

Word on the superhighway is there's a new version of OMeta coming, but I don't know that it will be based on Squeak (the most recent was built atop Javascript.) More news when I know for sure and have blessing to share with the group.

> On May 19, 2015, at 7:00 PM, "Phil (list)" <[hidden email]> wrote:
>
>> On Tue, 2015-05-19 at 19:12 -0300, Edgar J. De Cleene wrote:
>> Your info is very useful.
>> The experiment until now is I can load OMeta in Squeak 4.6 from
>> SqueakSource.
>> In my fork wich is Cuis 3,2 can’t load and several tricks for having
>> the classes ordered before OMeta uses each failed.
>> So if you have the original .st of Alessandro , wish have it for try.
>> In Cuis as cloned from Github the preload package loads fine and the
>> examples works as in Squeak 4.6.
>> I send mail here with link to .jpg picture.
>> But post load fails.
>
> OK, I've almost got it working again except for one tiny thing (that
> breaks everything)... #compilerClass no longer exists.  I've poked
> around and all of the compiler indirection has been removed (i.e.
> Compiler is hard-coded everywhere).  This completely breaks OMeta2 so
> unless Juan would consider bringing this functionality back, it's
> probably not going to work without essentially a rewrite.
>
> ... so that's why I gave up on it :-(
>
>>
>> Edgar
>>
>>
>> --
>> Edgar J. De Cleene
>> Sent with Sparrow
>>
>>
>>> On Tuesday, May 19, 2015 at 6:55 PM, Phil (list) wrote:
>>>
>>>> On Sat, 2015-05-16 at 05:41 -0300, Edgar J. De Cleene wrote:
>>>> People:
>>>>
>>>>
>>>> I working in my SqueakRosCore fork and think if Xtreams should be
>>>> on it or
>>>> not.
>>>>
>>>>
>>>> Google surfing brings me
>>>>
>>>>
>>>> http://comments.gmane.org/gmane.comp.lang.smalltalk.squeak.general/156355
>>>>
>>>>
>>>> So, four years later I wish feedback about this.
>>>>
>>>>
>>>> Cool working examples if you know of any.
>>>>
>>>>
>>>> Opinions:
>>>>
>>>>
>>>> SqueakSource have two projects about Xtreams, which one I should
>>>> have ?
>>>>
>>>>
>>>> Ometa pay same day the work for having it into the .image ?
>>>
>>>
>>> OK, it's all coming back to me as to why my port of OMeta2 died: I
>>> had
>>> it working perfectly in Cuis 1.0. I think I also had it working in
>>> Cuis
>>> 2.0 (I even have a .pck which I *believe* was kinda, sorta had
>>> working
>>> at one point... but now it can only be read in, it no longer works).
>>>
>>>
>>> IIRC, the issue was that part of OMeta is written in OMeta so you
>>> can't
>>> just load it in the way you would a regular package. I think
>>> original
>>> work by Alessandro was a .st file which had a very specific ordering
>>> of
>>> code to load and ran some code as part of the filein which installed
>>> OMeta into the image. The SqueakSource package addressed the loading
>>> issues by splitting it in 2: a preload and postload package. I
>>> believe
>>> the preload package patches the parser so that the postload package
>>> can
>>> even be parsed by Squeak. Then there is some code that gets called
>>> after the postload package has been read in that actually patches
>>> OMeta
>>> into the Squeak compiler. It's a bit tricky and changes to the
>>> Smalltalk parser/compiler since then will require some changes (I
>>> believe it broke as a result of the changes for closure support
>>> being
>>> added ~2012 or so)
>>>
>>>
>>> I'm not sure how far you've gotten with it on your own already... if
>>> what I have might be helpful, let me know and I'll see if I can't
>>> clean
>>> it up a bit and put it up on github.
>>>
>>>
>>>>
>>>>
>>>> Edgar
>>>> @morplenauta en twitter
>>>
>>>
>>> Thanks,
>>> Phil
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> Cuis mailing list
>>> [hidden email]
>>> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
>>
>>
>> _______________________________________________
>> Cuis mailing list
>> [hidden email]
>> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
>
>
>
> _______________________________________________
> Cuis mailing list
> [hidden email]
> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org

_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Phil B
On Tue, 2015-05-19 at 22:12 -0700, Casey Ransberger wrote:
> Hi Phil.
>
> Any particular reason compiler indirection can't go into one of our Squeak compat packs?
>

I'll let someone else volunteer for that headache. (see below)

> Got myself injured, and will be a week or three before I can sit at a computer for an extended period of time, but I could have a look at it if there isn't some reason we can't get it into compat. I surfed all the way through Cuis looking at every single reference to Color once. Seems like doing the same wouldn't be so hard looking for refs to Compiler and fixing them as I went.
>

Sorry to hear that and hope you get well soon.  Now would be a good time
for a tablet if you don't already have one...

> I guess the question is, can a pck modify core classes?
>

It could, but not terribly efficiently.  We're not talking about missing
methods, we're talking about needing to alter existing rather critical
ones.  A package can override those methods (from a quick glance it
looked to be well over a dozen) but it's an ongoing maintenance headache
as you would have to merge the needed code with what's already there.
This was the reason that #compilerClass existed: to provide an easy way
to swap out the compiler at whatever level is desired without having to
change numerous other methods and keeping those changes in sync with
upstream changes over time.

> --C
>
> P.S.
>
> Word on the superhighway is there's a new version of OMeta coming, but I don't know that it will be based on Squeak (the most recent was built atop Javascript.) More news when I know for sure and have blessing to share with the group.
>
> > On May 19, 2015, at 7:00 PM, "Phil (list)" <[hidden email]> wrote:
> >
> >> On Tue, 2015-05-19 at 19:12 -0300, Edgar J. De Cleene wrote:
> >> Your info is very useful.
> >> The experiment until now is I can load OMeta in Squeak 4.6 from
> >> SqueakSource.
> >> In my fork wich is Cuis 3,2 can’t load and several tricks for having
> >> the classes ordered before OMeta uses each failed.
> >> So if you have the original .st of Alessandro , wish have it for try.
> >> In Cuis as cloned from Github the preload package loads fine and the
> >> examples works as in Squeak 4.6.
> >> I send mail here with link to .jpg picture.
> >> But post load fails.
> >
> > OK, I've almost got it working again except for one tiny thing (that
> > breaks everything)... #compilerClass no longer exists.  I've poked
> > around and all of the compiler indirection has been removed (i.e.
> > Compiler is hard-coded everywhere).  This completely breaks OMeta2 so
> > unless Juan would consider bringing this functionality back, it's
> > probably not going to work without essentially a rewrite.
> >
> > ... so that's why I gave up on it :-(
> >
> >>
> >> Edgar
> >>
> >>
> >> --
> >> Edgar J. De Cleene
> >> Sent with Sparrow
> >>
> >>
> >>> On Tuesday, May 19, 2015 at 6:55 PM, Phil (list) wrote:
> >>>
> >>>> On Sat, 2015-05-16 at 05:41 -0300, Edgar J. De Cleene wrote:
> >>>> People:
> >>>>
> >>>>
> >>>> I working in my SqueakRosCore fork and think if Xtreams should be
> >>>> on it or
> >>>> not.
> >>>>
> >>>>
> >>>> Google surfing brings me
> >>>>
> >>>>
> >>>> http://comments.gmane.org/gmane.comp.lang.smalltalk.squeak.general/156355
> >>>>
> >>>>
> >>>> So, four years later I wish feedback about this.
> >>>>
> >>>>
> >>>> Cool working examples if you know of any.
> >>>>
> >>>>
> >>>> Opinions:
> >>>>
> >>>>
> >>>> SqueakSource have two projects about Xtreams, which one I should
> >>>> have ?
> >>>>
> >>>>
> >>>> Ometa pay same day the work for having it into the .image ?
> >>>
> >>>
> >>> OK, it's all coming back to me as to why my port of OMeta2 died: I
> >>> had
> >>> it working perfectly in Cuis 1.0. I think I also had it working in
> >>> Cuis
> >>> 2.0 (I even have a .pck which I *believe* was kinda, sorta had
> >>> working
> >>> at one point... but now it can only be read in, it no longer works).
> >>>
> >>>
> >>> IIRC, the issue was that part of OMeta is written in OMeta so you
> >>> can't
> >>> just load it in the way you would a regular package. I think
> >>> original
> >>> work by Alessandro was a .st file which had a very specific ordering
> >>> of
> >>> code to load and ran some code as part of the filein which installed
> >>> OMeta into the image. The SqueakSource package addressed the loading
> >>> issues by splitting it in 2: a preload and postload package. I
> >>> believe
> >>> the preload package patches the parser so that the postload package
> >>> can
> >>> even be parsed by Squeak. Then there is some code that gets called
> >>> after the postload package has been read in that actually patches
> >>> OMeta
> >>> into the Squeak compiler. It's a bit tricky and changes to the
> >>> Smalltalk parser/compiler since then will require some changes (I
> >>> believe it broke as a result of the changes for closure support
> >>> being
> >>> added ~2012 or so)
> >>>
> >>>
> >>> I'm not sure how far you've gotten with it on your own already... if
> >>> what I have might be helpful, let me know and I'll see if I can't
> >>> clean
> >>> it up a bit and put it up on github.
> >>>
> >>>
> >>>>
> >>>>
> >>>> Edgar
> >>>> @morplenauta en twitter
> >>>
> >>>
> >>> Thanks,
> >>> Phil
> >>>
> >>>
> >>>
> >>>
> >>> _______________________________________________
> >>> Cuis mailing list
> >>> [hidden email]
> >>> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
> >>
> >>
> >> _______________________________________________
> >> Cuis mailing list
> >> [hidden email]
> >> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
> >
> >
> >
> > _______________________________________________
> > Cuis mailing list
> > [hidden email]
> > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
>
> _______________________________________________
> Cuis mailing list
> [hidden email]
> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org



_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Casey Ransberger-2
Are you arguing for restoring compiler indirection in Cuis?

That's a trade off either way. But I agree having it in the core of the system would be worthwhile for language researchers.

Is hosting them worth it to us as a community? I'd argue that it is, but that's me. It has to serve the whole community, or a large subset, before it is worth incorporating.

--C

> On May 19, 2015, at 11:36 PM, "Phil (list)" <[hidden email]> wrote:
>
>> On Tue, 2015-05-19 at 22:12 -0700, Casey Ransberger wrote:
>> Hi Phil.
>>
>> Any particular reason compiler indirection can't go into one of our Squeak compat packs?
>
> I'll let someone else volunteer for that headache. (see below)
>
>> Got myself injured, and will be a week or three before I can sit at a computer for an extended period of time, but I could have a look at it if there isn't some reason we can't get it into compat. I surfed all the way through Cuis looking at every single reference to Color once. Seems like doing the same wouldn't be so hard looking for refs to Compiler and fixing them as I went.
>
> Sorry to hear that and hope you get well soon.  Now would be a good time
> for a tablet if you don't already have one...
>
>> I guess the question is, can a pck modify core classes?
>
> It could, but not terribly efficiently.  We're not talking about missing
> methods, we're talking about needing to alter existing rather critical
> ones.  A package can override those methods (from a quick glance it
> looked to be well over a dozen) but it's an ongoing maintenance headache
> as you would have to merge the needed code with what's already there.
> This was the reason that #compilerClass existed: to provide an easy way
> to swap out the compiler at whatever level is desired without having to
> change numerous other methods and keeping those changes in sync with
> upstream changes over time.
>
>> --C
>>
>> P.S.
>>
>> Word on the superhighway is there's a new version of OMeta coming, but I don't know that it will be based on Squeak (the most recent was built atop Javascript.) More news when I know for sure and have blessing to share with the group.
>>
>>>> On May 19, 2015, at 7:00 PM, "Phil (list)" <[hidden email]> wrote:
>>>>
>>>> On Tue, 2015-05-19 at 19:12 -0300, Edgar J. De Cleene wrote:
>>>> Your info is very useful.
>>>> The experiment until now is I can load OMeta in Squeak 4.6 from
>>>> SqueakSource.
>>>> In my fork wich is Cuis 3,2 can’t load and several tricks for having
>>>> the classes ordered before OMeta uses each failed.
>>>> So if you have the original .st of Alessandro , wish have it for try.
>>>> In Cuis as cloned from Github the preload package loads fine and the
>>>> examples works as in Squeak 4.6.
>>>> I send mail here with link to .jpg picture.
>>>> But post load fails.
>>>
>>> OK, I've almost got it working again except for one tiny thing (that
>>> breaks everything)... #compilerClass no longer exists.  I've poked
>>> around and all of the compiler indirection has been removed (i.e.
>>> Compiler is hard-coded everywhere).  This completely breaks OMeta2 so
>>> unless Juan would consider bringing this functionality back, it's
>>> probably not going to work without essentially a rewrite.
>>>
>>> ... so that's why I gave up on it :-(
>>>
>>>>
>>>> Edgar
>>>>
>>>>
>>>> --
>>>> Edgar J. De Cleene
>>>> Sent with Sparrow
>>>>
>>>>
>>>>>> On Tuesday, May 19, 2015 at 6:55 PM, Phil (list) wrote:
>>>>>>
>>>>>> On Sat, 2015-05-16 at 05:41 -0300, Edgar J. De Cleene wrote:
>>>>>> People:
>>>>>>
>>>>>>
>>>>>> I working in my SqueakRosCore fork and think if Xtreams should be
>>>>>> on it or
>>>>>> not.
>>>>>>
>>>>>>
>>>>>> Google surfing brings me
>>>>>>
>>>>>>
>>>>>> http://comments.gmane.org/gmane.comp.lang.smalltalk.squeak.general/156355
>>>>>>
>>>>>>
>>>>>> So, four years later I wish feedback about this.
>>>>>>
>>>>>>
>>>>>> Cool working examples if you know of any.
>>>>>>
>>>>>>
>>>>>> Opinions:
>>>>>>
>>>>>>
>>>>>> SqueakSource have two projects about Xtreams, which one I should
>>>>>> have ?
>>>>>>
>>>>>>
>>>>>> Ometa pay same day the work for having it into the .image ?
>>>>>
>>>>>
>>>>> OK, it's all coming back to me as to why my port of OMeta2 died: I
>>>>> had
>>>>> it working perfectly in Cuis 1.0. I think I also had it working in
>>>>> Cuis
>>>>> 2.0 (I even have a .pck which I *believe* was kinda, sorta had
>>>>> working
>>>>> at one point... but now it can only be read in, it no longer works).
>>>>>
>>>>>
>>>>> IIRC, the issue was that part of OMeta is written in OMeta so you
>>>>> can't
>>>>> just load it in the way you would a regular package. I think
>>>>> original
>>>>> work by Alessandro was a .st file which had a very specific ordering
>>>>> of
>>>>> code to load and ran some code as part of the filein which installed
>>>>> OMeta into the image. The SqueakSource package addressed the loading
>>>>> issues by splitting it in 2: a preload and postload package. I
>>>>> believe
>>>>> the preload package patches the parser so that the postload package
>>>>> can
>>>>> even be parsed by Squeak. Then there is some code that gets called
>>>>> after the postload package has been read in that actually patches
>>>>> OMeta
>>>>> into the Squeak compiler. It's a bit tricky and changes to the
>>>>> Smalltalk parser/compiler since then will require some changes (I
>>>>> believe it broke as a result of the changes for closure support
>>>>> being
>>>>> added ~2012 or so)
>>>>>
>>>>>
>>>>> I'm not sure how far you've gotten with it on your own already... if
>>>>> what I have might be helpful, let me know and I'll see if I can't
>>>>> clean
>>>>> it up a bit and put it up on github.
>>>>>
>>>>>
>>>>>>
>>>>>>
>>>>>> Edgar
>>>>>> @morplenauta en twitter
>>>>>
>>>>>
>>>>> Thanks,
>>>>> Phil
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Cuis mailing list
>>>>> [hidden email]
>>>>> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
>>>>
>>>>
>>>> _______________________________________________
>>>> Cuis mailing list
>>>> [hidden email]
>>>> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
>>>
>>>
>>>
>>> _______________________________________________
>>> Cuis mailing list
>>> [hidden email]
>>> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
>>
>> _______________________________________________
>> Cuis mailing list
>> [hidden email]
>> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
>
>
>
> _______________________________________________
> Cuis mailing list
> [hidden email]
> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org

_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Phil B
On Wed, 2015-05-20 at 00:02 -0700, Casey Ransberger wrote:
> Are you arguing for restoring compiler indirection in Cuis?
>
> That's a trade off either way. But I agree having it in the core of the system would be worthwhile for language researchers.
>

I'm in favor of it but I may be in the minority on this.  If feels like
a trade off of extensibility in favor of minimalism.

> Is hosting them worth it to us as a community? I'd argue that it is, but that's me. It has to serve the whole community, or a large subset, before it is worth incorporating.
>

I think so.  Granted there are only a small number of use cases for it,
but for those cases I can't think of a substitute for the capability it
provided... extensibility is hard to add via a filein without being
brittle.

> --C
>
> > On May 19, 2015, at 11:36 PM, "Phil (list)" <[hidden email]> wrote:
> >
> >> On Tue, 2015-05-19 at 22:12 -0700, Casey Ransberger wrote:
> >> Hi Phil.
> >>
> >> Any particular reason compiler indirection can't go into one of our Squeak compat packs?
> >
> > I'll let someone else volunteer for that headache. (see below)
> >
> >> Got myself injured, and will be a week or three before I can sit at a computer for an extended period of time, but I could have a look at it if there isn't some reason we can't get it into compat. I surfed all the way through Cuis looking at every single reference to Color once. Seems like doing the same wouldn't be so hard looking for refs to Compiler and fixing them as I went.
> >
> > Sorry to hear that and hope you get well soon.  Now would be a good time
> > for a tablet if you don't already have one...
> >
> >> I guess the question is, can a pck modify core classes?
> >
> > It could, but not terribly efficiently.  We're not talking about missing
> > methods, we're talking about needing to alter existing rather critical
> > ones.  A package can override those methods (from a quick glance it
> > looked to be well over a dozen) but it's an ongoing maintenance headache
> > as you would have to merge the needed code with what's already there.
> > This was the reason that #compilerClass existed: to provide an easy way
> > to swap out the compiler at whatever level is desired without having to
> > change numerous other methods and keeping those changes in sync with
> > upstream changes over time.
> >
> >> --C
> >>
> >> P.S.
> >>
> >> Word on the superhighway is there's a new version of OMeta coming, but I don't know that it will be based on Squeak (the most recent was built atop Javascript.) More news when I know for sure and have blessing to share with the group.
> >>
> >>>> On May 19, 2015, at 7:00 PM, "Phil (list)" <[hidden email]> wrote:
> >>>>
> >>>> On Tue, 2015-05-19 at 19:12 -0300, Edgar J. De Cleene wrote:
> >>>> Your info is very useful.
> >>>> The experiment until now is I can load OMeta in Squeak 4.6 from
> >>>> SqueakSource.
> >>>> In my fork wich is Cuis 3,2 can’t load and several tricks for having
> >>>> the classes ordered before OMeta uses each failed.
> >>>> So if you have the original .st of Alessandro , wish have it for try.
> >>>> In Cuis as cloned from Github the preload package loads fine and the
> >>>> examples works as in Squeak 4.6.
> >>>> I send mail here with link to .jpg picture.
> >>>> But post load fails.
> >>>
> >>> OK, I've almost got it working again except for one tiny thing (that
> >>> breaks everything)... #compilerClass no longer exists.  I've poked
> >>> around and all of the compiler indirection has been removed (i.e.
> >>> Compiler is hard-coded everywhere).  This completely breaks OMeta2 so
> >>> unless Juan would consider bringing this functionality back, it's
> >>> probably not going to work without essentially a rewrite.
> >>>
> >>> ... so that's why I gave up on it :-(
> >>>
> >>>>
> >>>> Edgar
> >>>>
> >>>>
> >>>> --
> >>>> Edgar J. De Cleene
> >>>> Sent with Sparrow
> >>>>
> >>>>
> >>>>>> On Tuesday, May 19, 2015 at 6:55 PM, Phil (list) wrote:
> >>>>>>
> >>>>>> On Sat, 2015-05-16 at 05:41 -0300, Edgar J. De Cleene wrote:
> >>>>>> People:
> >>>>>>
> >>>>>>
> >>>>>> I working in my SqueakRosCore fork and think if Xtreams should be
> >>>>>> on it or
> >>>>>> not.
> >>>>>>
> >>>>>>
> >>>>>> Google surfing brings me
> >>>>>>
> >>>>>>
> >>>>>> http://comments.gmane.org/gmane.comp.lang.smalltalk.squeak.general/156355
> >>>>>>
> >>>>>>
> >>>>>> So, four years later I wish feedback about this.
> >>>>>>
> >>>>>>
> >>>>>> Cool working examples if you know of any.
> >>>>>>
> >>>>>>
> >>>>>> Opinions:
> >>>>>>
> >>>>>>
> >>>>>> SqueakSource have two projects about Xtreams, which one I should
> >>>>>> have ?
> >>>>>>
> >>>>>>
> >>>>>> Ometa pay same day the work for having it into the .image ?
> >>>>>
> >>>>>
> >>>>> OK, it's all coming back to me as to why my port of OMeta2 died: I
> >>>>> had
> >>>>> it working perfectly in Cuis 1.0. I think I also had it working in
> >>>>> Cuis
> >>>>> 2.0 (I even have a .pck which I *believe* was kinda, sorta had
> >>>>> working
> >>>>> at one point... but now it can only be read in, it no longer works).
> >>>>>
> >>>>>
> >>>>> IIRC, the issue was that part of OMeta is written in OMeta so you
> >>>>> can't
> >>>>> just load it in the way you would a regular package. I think
> >>>>> original
> >>>>> work by Alessandro was a .st file which had a very specific ordering
> >>>>> of
> >>>>> code to load and ran some code as part of the filein which installed
> >>>>> OMeta into the image. The SqueakSource package addressed the loading
> >>>>> issues by splitting it in 2: a preload and postload package. I
> >>>>> believe
> >>>>> the preload package patches the parser so that the postload package
> >>>>> can
> >>>>> even be parsed by Squeak. Then there is some code that gets called
> >>>>> after the postload package has been read in that actually patches
> >>>>> OMeta
> >>>>> into the Squeak compiler. It's a bit tricky and changes to the
> >>>>> Smalltalk parser/compiler since then will require some changes (I
> >>>>> believe it broke as a result of the changes for closure support
> >>>>> being
> >>>>> added ~2012 or so)
> >>>>>
> >>>>>
> >>>>> I'm not sure how far you've gotten with it on your own already... if
> >>>>> what I have might be helpful, let me know and I'll see if I can't
> >>>>> clean
> >>>>> it up a bit and put it up on github.
> >>>>>
> >>>>>
> >>>>>>
> >>>>>>
> >>>>>> Edgar
> >>>>>> @morplenauta en twitter
> >>>>>
> >>>>>
> >>>>> Thanks,
> >>>>> Phil
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>> _______________________________________________
> >>>>> Cuis mailing list
> >>>>> [hidden email]
> >>>>> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
> >>>>
> >>>>
> >>>> _______________________________________________
> >>>> Cuis mailing list
> >>>> [hidden email]
> >>>> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
> >>>
> >>>
> >>>
> >>> _______________________________________________
> >>> Cuis mailing list
> >>> [hidden email]
> >>> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
> >>
> >> _______________________________________________
> >> Cuis mailing list
> >> [hidden email]
> >> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
> >
> >
> >
> > _______________________________________________
> > Cuis mailing list
> > [hidden email]
> > http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
>
> _______________________________________________
> Cuis mailing list
> [hidden email]
> http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org



_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Juan Vuletich-4
In reply to this post by Phil B
On 5/19/2015 11:00 PM, Phil (list) wrote:

> On Tue, 2015-05-19 at 19:12 -0300, Edgar J. De Cleene wrote:
>> Your info is very useful.
>> The experiment until now is I can load OMeta in Squeak 4.6 from
>> SqueakSource.
>> In my fork wich is Cuis 3,2 can’t load and several tricks for having
>> the classes ordered before OMeta uses each failed.
>> So if you have the original .st of Alessandro , wish have it for try.
>> In Cuis as cloned from Github the preload package loads fine and the
>> examples works as in Squeak 4.6.
>> I send mail here with link to .jpg picture.
>> But post load fails.
>>
> OK, I've almost got it working again except for one tiny thing (that
> breaks everything)... #compilerClass no longer exists.  I've poked
> around and all of the compiler indirection has been removed (i.e.
> Compiler is hard-coded everywhere).  This completely breaks OMeta2 so
> unless Juan would consider bringing this functionality back, it's
> probably not going to work without essentially a rewrite.
>
> ... so that's why I gave up on it :-(

Ok. Let me take a look.

Cheers,
Juan Vuletich

_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Juan Vuletich-4
In reply to this post by Phil B
Hi Folks,

On 5/20/2015 4:37 AM, Phil (list) wrote:
> On Wed, 2015-05-20 at 00:02 -0700, Casey Ransberger wrote:
>> Are you arguing for restoring compiler indirection in Cuis?
>>
>> That's a trade off either way. But I agree having it in the core of the system would be worthwhile for language researchers.
>>
> I'm in favor of it but I may be in the minority on this.  If feels like
> a trade off of extensibility in favor of minimalism.

Minimalism should be to empower us, not to impair us. Removing this was
a mistake.

Please try the attached. If it's ok, it will be in the next commit.

Cheers,
Juan Vuletich

_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org

compilerClass.zip (24K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Phil B
Juan,


On Wed, 2015-05-20 at 13:11 -0300, Juan Vuletich wrote:

> Hi Folks,
>
> On 5/20/2015 4:37 AM, Phil (list) wrote:
> > On Wed, 2015-05-20 at 00:02 -0700, Casey Ransberger wrote:
> >> Are you arguing for restoring compiler indirection in Cuis?
> >>
> >> That's a trade off either way. But I agree having it in the core of the system would be worthwhile for language researchers.
> >>
> > I'm in favor of it but I may be in the minority on this.  If feels like
> > a trade off of extensibility in favor of minimalism.
>
> Minimalism should be to empower us, not to impair us. Removing this was
> a mistake.
>
> Please try the attached. If it's ok, it will be in the next commit.
>

This is fantastic news!  I tried the changesets against 2330 but there
still appear to be some direct calls left (around 2 dozen... filein
failed on PositionableStream #fileInAnnouncing:)  I *believe* pretty
much every reference to Compiler needs to be replaced with compilerClass
for class of the method being compiled/compiled for (i.e. the caller).

> Cheers,
> Juan Vuletich

Thanks,
Phil


_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
Reply | Threaded
Open this post in threaded view
|

Re: Xtreams , Ometa and related

Juan Vuletich-4
Hi Phil,

On Wed, May 20, 2015 2:26 pm, Phil (list) wrote:

> Juan,
>
>
>
> On Wed, 2015-05-20 at 13:11 -0300, Juan Vuletich wrote:
>
>> Hi Folks,
>>
>>
>> On 5/20/2015 4:37 AM, Phil (list) wrote:
>>
>>> On Wed, 2015-05-20 at 00:02 -0700, Casey Ransberger wrote:
>>>
>>>> Are you arguing for restoring compiler indirection in Cuis?
>>>>
>>>>
>>>> That's a trade off either way. But I agree having it in the core of
>>>> the system would be worthwhile for language researchers.
>>>>
>>> I'm in favor of it but I may be in the minority on this.  If feels
>>> like a trade off of extensibility in favor of minimalism.
>>
>> Minimalism should be to empower us, not to impair us. Removing this was
>>  a mistake.
>>
>> Please try the attached. If it's ok, it will be in the next commit.
>>
>>
>
> This is fantastic news!  I tried the changesets against 2330 but there
> still appear to be some direct calls left (around 2 dozen... filein failed
> on PositionableStream #fileInAnnouncing:)  I *believe* pretty much every
> reference to Compiler needs to be replaced with compilerClass for class of
> the method being compiled/compiled for (i.e. the caller).
>
>> Cheers,
>> Juan Vuletich
>>
>
> Thanks,
> Phil

I did as follow:
1) start Cuis 2330
2) Open file list
3) click on 2331
4) click on [install]
5) click on 2332
6) click on [install]
and got no error messages. Please detail what you did, and what error you
get.

The references to Compiler that are left are pretty much the same as in
Squeak 4.5. Which of those gives you problems?

Thanks,
Juan Vuletich

_______________________________________________
Cuis mailing list
[hidden email]
http://jvuletich.org/mailman/listinfo/cuis_jvuletich.org
12