Nabble - Squeak - Dev

Re: [squeak-dev] MNU after recent trunk update

2009-10-09T06:31:54Z

A in transit initialization was missing. The update process should
work again. Sorry for the inconvenience.

Alex

2009/10/9 Vaidotas Didžbalis <vaidasd@...>:
> Log is attached,
> Vaidotas
>
>
>
>

[squeak-dev] The Trunk: Morphic-laza.209.mcz

2009-10-09T06:23:42Z

Alexander Lazarević uploaded a new version of Morphic to project The Trunk:
http://source.squeak.org/trunk/Morphic-laza.209.mcz

==================== Summary ====================

Name: Morphic-laza.209
Author: laza
Time: 9 October 2009, 3:22:49 am
UUID: c8f72563-e1f9-fc45-86cc-e8aa4bcdb2b9
Ancestors: Morphic-laza.208

Remove the in transition initialization again

=============== Diff against Morphic-laza.208 ===============

Item was changed:
----- Method: SystemProgressMorph>>recenter (in category 'private') -----
recenter
| position |
- self class initialize.
"Put ourself in the center of the display"
self align: self fullBounds center with: Display boundingBox center.
"Check to see if labels are wider than progress bars. In that case do
a centered instead of the default left aligned layout."
position := self width > (Inset x * 2 + (self borderWidth * 2) + BarWidth)
ifTrue: [#topCenter]
ifFalse: [#leftCenter].
self cellPositioning: position!

Item was changed:
----- Method: SystemProgressMorph>>freeSlot: (in category 'private') -----
freeSlot: number
number > 0 ifTrue: [
lock critical: [
(bars at: number) delete.
(labels at: number) delete.
activeSlots := activeSlots - 1.
activeSlots = 0
+ ifTrue: [self delete]
- ifTrue: [self delete. self class initialize]
ifFalse: [self recenter]]]!

[squeak-dev] The Trunk: Morphic-laza.208.mcz

2009-10-09T05:54:00Z

Alexander Lazarević uploaded a new version of Morphic to project The Trunk:
http://source.squeak.org/trunk/Morphic-laza.208.mcz

==================== Summary ====================

Name: Morphic-laza.208
Author: laza
Time: 9 October 2009, 2:53 am
UUID: 535a689a-d7ae-654a-b1ab-984d3f51f560
Ancestors: Morphic-laza.207

Make sure SystemProgressMorph gets initialized during update process

=============== Diff against Morphic-laza.207 ===============

Item was changed:
----- Method: SystemProgressMorph>>recenter (in category 'private') -----
recenter
| position |
+ self class initialize.
"Put ourself in the center of the display"
self align: self fullBounds center with: Display boundingBox center.
"Check to see if labels are wider than progress bars. In that case do
a centered instead of the default left aligned layout."
position := self width > (Inset x * 2 + (self borderWidth * 2) + BarWidth)
ifTrue: [#topCenter]
ifFalse: [#leftCenter].
self cellPositioning: position!

Item was changed:
----- Method: SystemProgressMorph>>freeSlot: (in category 'private') -----
freeSlot: number
number > 0 ifTrue: [
lock critical: [
(bars at: number) delete.
(labels at: number) delete.
activeSlots := activeSlots - 1.
activeSlots = 0
+ ifTrue: [self delete. self class initialize]
- ifTrue: [self delete]
ifFalse: [self recenter]]]!

Item was changed:

[squeak-dev] The Trunk: Morphic-laza.207.mcz

2009-10-09T05:40:37Z

Alexander Lazarević uploaded a new version of Morphic to project The Trunk:
http://source.squeak.org/trunk/Morphic-laza.207.mcz

==================== Summary ====================

Name: Morphic-laza.207
Author: laza
Time: 9 October 2009, 2:39:14 am
UUID: c39f5ea2-dc5a-504d-b6e2-048f544de535
Ancestors: Morphic-laza.206

Initialize SystemProgressMorph after loading the package

=============== Diff against Morphic-laza.206 ===============

Item was changed:

[squeak-dev] MNU after recent trunk update

2009-10-09T04:53:51Z

Log is attached,
Vaidotas

SqueakDebug.log (5K) Download Attachment

[squeak-dev] The Trunk: Morphic-laza.206.mcz

2009-10-09T02:46:49Z

Alexander Lazarević uploaded a new version of Morphic to project The Trunk:
http://source.squeak.org/trunk/Morphic-laza.206.mcz

==================== Summary ====================

Name: Morphic-laza.206
Author: laza
Time: 9 October 2009, 11:45:50 am
UUID: 0e18421c-498b-9946-88cb-c21791ae921e
Ancestors: Morphic-MAD.205

Use a centered layout if labels are too long

=============== Diff against Morphic-MAD.205 ===============

Item was changed:
----- Method: SystemProgressMorph class>>initialize (in category 'class initialization') -----
initialize
"SystemProgressMorph initialize; reset"
BarHeight := 8.
+ BarWidth := 300.
+ Inset := 30@30!
- BarWidth := 300!

Item was changed:
RectangleMorph subclass: #SystemProgressMorph
instanceVariableNames: 'activeSlots bars labels font lock'
+ classVariableNames: 'BarHeight BarWidth Inset UniqueInstance'
- classVariableNames: 'BarHeight BarWidth UniqueInstance'
poolDictionaries: ''
category: 'Morphic-Widgets'!

!SystemProgressMorph commentStamp: '<historical>' prior: 0!
An single instance of this morph class is used to display progress while the system is busy, eg. while it receives code updates or does a fileIn. To give the user progress information you don't deal directly with SystemProgressMorph. You keep on using the well established way of progress notification, that has been a long time in the system, is widely used and does not depend on the existence of SystemProgressMorph. For more information on this look at the example in this class or look at the comment of the method displayProgressAt:from:to:during: in class String.

SystemProgressMorph is not meant to be used as a component inside other morphs.

You can switch back to the old style of progress display by disabling the morphicProgressStyle setting in the morphic section of the preferences.!

Item was changed:
----- Method: SystemProgressMorph>>recenter (in category 'private') -----
recenter
+ | position |
+ "Put ourself in the center of the display"
self align: self fullBounds center with: Display boundingBox center.
+ "Check to see if labels are wider than progress bars. In that case do
+ a centered instead of the default left aligned layout."
+ position := self width > (Inset x * 2 + (self borderWidth * 2) + BarWidth)
+ ifTrue: [#topCenter]
+ ifFalse: [#leftCenter].
+ self cellPositioning: position!
- !

Item was changed:
----- Method: SystemProgressMorph>>initialize (in category 'initialization') -----
initialize
super initialize.
activeSlots := 0.
bars := Array new: 10.
labels := Array new: 10.
font := Preferences standardMenuFont.
lock := Semaphore forMutualExclusion.
self setDefaultParameters;
setProperty: #morphicLayerNumber toValue: self morphicLayerNumber;
layoutPolicy: TableLayout new;
listDirection: #topToBottom;
cellPositioning: #leftCenter;
cellInset: 5;
listCentering: #center;
hResizing: #shrinkWrap;
vResizing: #shrinkWrap;
+ layoutInset: Inset;
- layoutInset:30@30;
minWidth: 150!

[squeak-dev] !! Let us know you're listening

2009-10-08T22:27:16Z

Hello,

If you're reading this message, and you like Etoys, please take a minute and let us know you're listening:

http://squeakland.org/squeakers/

We're trying to get a better idea of the size of the Etoys community.

Also, please forward the link to other Squeakers that you know.

Thank you,

Timothy

Timothy Falconer

Squeakland Foundation

http://squeakland.org

610-797-3100

"Intelligence is what you use when you don't know what to do." ... piaget

Re: [squeak-dev] Re: Selection in new file list

2009-10-08T19:46:44Z

On Thursday 08 October 2009 09:34:41 am Andreas Raab wrote:
> Perhaps a binary operator #/ would be in order? I.e.,
>
> foobar := FileDirectory default / 'foo' / 'bar'.
>
> (although this overloads the meaning of #/ which is not great either)
It is an excellent idea if URI extends Collection. Collections like
squeakland, temp directory etc. could be defined as separate name spaces like
home://, tmp://, squeakland:// and map entries to pathnames at run time.

Imagine begin able to load projects with:
squeakland := URI namespace: 'squeakland'.
project = squeakland / 'featured projects' at: 'brule.pr'.

Subbu

[squeak-dev] Re: Selection in new file list

2009-10-08T10:34:18Z

On 2009-10-07, at 9:53 PM, Andreas Raab wrote:

> John M McIntosh wrote:
>> Well we used URI with great success in Sophie, since we could use a
>> relative path then resolve it to a network path, file system path,
>> or a specialized in memory path in order to store book objects in
>> those
>> three storage areas.
>> But a few things come to mind for example we could someone do
>> foo := CPlatform homeDirectory or documentDirectory or
>> tempDirectory or musicDirectory and it would mostly do the right
>> thing based on the platform to construct a URI.
>
> Interesting. The result was a URI instead of a FileDirectory? In
> Qwaq^H^H^H^H Teleplace we we use e.g., FileDirectory [temp|home|log|
> cache]Directory to return instances of FileDirectory. Or did you
> implement most of the interesting file creation methods on URIs as
> well?

As the Sophie storage Czar I ensured everything was a URI. The book
storage manager then would instantiate a concrete class that would
resolve the URI to the particular target (file, network, memory) at the
point of doing the open/read/write/delete/copy.

>
>> then what if you did
>> fum := CPlatform homeDirectory , 'myStuff' , 'calledWhat' ,
>> 'where.txt' .
>
> Ooooh. Comma. I like that. Already a generic concatenation operator.
> Much better than slash.
>
>> using the ',' operator to concatenate a path together. Then a
>> method to return the proper path based on the type you want, and of
>> course the platform specific details.
>> platformDirectoryPath = fum asFileSytemDirectoryPath.
>
> That is quite nice. Yeah I could see using that. Any chance you can
> re-release the URI code as MIT? (IIRC, Sophie is BSD, right?).

Oooh a license question. Well you would have to ask Michael, I'd
*believe* the majority of the URI code predated usage in Sophie.

However maybe we want to consider the issue, where this path to an
object and then why would we make the object a URI or a text string?
Perhaps it should be it's own unique class which you resolve to a URI
or filesystempath when needed.

As mentioned in another note the issue(s) for example

path := path,'foo' does that mean we want to add the foo directory
or tack on 'foo' to the last textual item?

>
> Cheers,
> - Andreas

--
=
=
=
========================================================================
John M. McIntosh <johnmci@...> Twitter:
squeaker68882
Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com
=
=
=
========================================================================

Re: [squeak-dev] Installing seaside in squeak trunk image

2009-10-08T08:09:59Z

Andreas instructions worked fine, thank you all for your help.

I have a question, though. What should I use to install packages? Universe? Installer?

On Thu, Oct 8, 2009 at 3:03 AM, radoslav hodnicak <rh@...> wrote:

All you need is a recent version of the package DynamicBindings from http://www.squeaksource.com/KomHttpServer.html

On Thu, 8 Oct 2009, Ricardo Moran wrote:

Hi, I just installed Seaside version 2.8.4.591 from Universe. It finished ok
but when I tried to start the server with
WAKomEncoded startOn: 8080.

I got BlockClosure>>doesNotUnderstand: #newProcessWithBindings:

I searched a Seaside One click experience image (which is made with Pharo)
and I found this method in *DynamicBindings category of BlockClosure. In
order to make it work I simply filed out the whole category and loaded it in
squeak. It seems to be working but I wanted to make sure I was doing it
right. The thing is I want to show seaside to a friend that doesn't know
anything about smalltalk so I don't want him to get frustrated with obscure
bugs because of a bad installation. I suppose it would be best if I just
give him the Seaside one click experience image but I really don't like
Pharo.
Also, I would like to know what other package do you think I should install
in order to make a good starting environment for him.
Thanks in advance.

Re: [squeak-dev] Re: Selection in new file list / file system api

2009-10-08T04:35:36Z

On Thu, Oct 8, 2009 at 2:03 AM, Torsten Bergmann <astares@...> wrote:
> Damien also proposed a very small API that allows you to browse
> the file system easily including tests, just check
>
> http://code.google.com/p/pharo/issues/detail?id=1146

Forget about that. This is a very *very* small API that provide the
very strict minimum to browse a file system.

--
Damien Cassou
http://damiencassou.seasidehosting.st

"Lambdas are relegated to relative obscurity until Java makes them
popular by not having them." James Iry

[squeak-dev] Extras repository

2009-10-08T02:07:23Z

On 08.10.2009, at 10:18, Ronald Spengler wrote:

> I think an extras repo would be cool. I'm in the midst of some
> experiments with factoring a skinning infrastructure out of Polymorph,
> and when I come up with the right approach, it won't look much like
> polymorph anymore (I could create a new project on Squeaksource.com)
> and won't be stable enough to go into a base image any time soon. One
> of the design decisions I've made is to throw out compat with forks,
> so it does make sense to track the work along side that of the trunk.
>
> Another good reason to have an extras repo is: people who are doing
> work on applications which don't belong in a base image will have the
> opportunity to participate in the process with the same tools that the
> trunk uses. So we'll feel the same pain with the outages, etc. This
> encourages people working on their own projects to contribute fixes to
> squeaksource, etc.

While I agree having an extras repo would be good, I wouldn't see it
as repository for applications. Those have a fine place at squeaksource.com
IMHO. "Extras" on source.squeak.org would rather contain stuff
specific to Squeak trunk, no?

- Bert -

Re: [squeak-dev] http://source.squeak.org/inbox emptied

2009-10-08T01:18:39Z

I think an extras repo would be cool. I'm in the midst of some
experiments with factoring a skinning infrastructure out of Polymorph,
and when I come up with the right approach, it won't look much like
polymorph anymore (I could create a new project on Squeaksource.com)
and won't be stable enough to go into a base image any time soon. One
of the design decisions I've made is to throw out compat with forks,
so it does make sense to track the work along side that of the trunk.

Another good reason to have an extras repo is: people who are doing
work on applications which don't belong in a base image will have the
opportunity to participate in the process with the same tools that the
trunk uses. So we'll feel the same pain with the outages, etc. This
encourages people working on their own projects to contribute fixes to
squeaksource, etc.

On Tuesday, October 6, 2009, Michael van der Gulik <mikevdg@...> wrote:

>
>
> On Tue, Oct 6, 2009 at 5:25 PM, Andreas Raab <andreas.raab@... <javascript:_e({}, 'cvml', 'andreas.raab@...');>> wrote:
>
>
>
> The good news is that the inbox is empty now and looking for more contributions. Keep 'em coming!
>
>
> Wow! Good work!
>
> Gulik.
> --
> http://gulik.pbwiki.com/
>

--
Ron

[squeak-dev] Re: Selection in new file list / file system api

2009-10-08T01:03:11Z

Colin wrote on squeak-dev:
>I've been working on a new filesystem API built on top of the
>FilePlugin primitives.
>...
>if anybody wants to take a look it's here:
>http://source.wiresong.ca/mc/Filesystem-cwp.22.mcz

Damien also proposed a very small API that allows you to browse
the file system easily including tests, just check

http://code.google.com/p/pharo/issues/detail?id=1146

Bye
Torsten
--
GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT!
Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01

Re: [squeak-dev] Re: Selection in new file list

2009-10-08T00:39:26Z

On 7-Oct-09, at 9:04 PM, Andreas Raab wrote:

> Perhaps a binary operator #/ would be in order? I.e.,
>
> foobar := FileDirectory default / 'foo' / 'bar'.

I wasn't actually going to mention this for a few more days, but since
it's come up...

I've been working on a new filesystem API built on top of the
FilePlugin primitives. It does just what you propose above.
The following gives an absolute path:

ref := FSReference defaultDirectory / 'foo' / 'bar'.

A reference is just well, a reference, to a file or directory, not an
actual open stream, so it can respond to

#exists
#isDirectory
#isFile
#delete
#copyTo:
#fileStreamDo:

and so on. It's much nicer than FileDirectory.

It's completely untested on Windows, and there's no documentation yet,
but if anybody wants to take a look it's here:

http://source.wiresong.ca/mc/Filesystem-cwp.22.mcz

Colin

Re: [squeak-dev] Re: Selection in new file list

2009-10-08T00:14:44Z

Andreas Raab wrote:
>> then what if you did
>>
>> fum := CPlatform homeDirectory , 'myStuff' , 'calledWhat' ,
>> 'where.txt' .
>
> Ooooh. Comma. I like that. Already a generic concatenation operator.
> Much better than slash.

i often build filenames using the comma (e.g. name, '.txt')
so i don't think that its good to use the comma
as a directory separator.

just my few 2c
thanks
wolfgang

Re: [squeak-dev] Installing seaside in squeak trunk image

2009-10-07T23:03:21Z

All you need is a recent version of the package DynamicBindings from
http://www.squeaksource.com/KomHttpServer.html

On Thu, 8 Oct 2009, Ricardo Moran wrote:

> Hi, I just installed Seaside version 2.8.4.591 from Universe. It finished ok
> but when I tried to start the server with
> WAKomEncoded startOn: 8080.
>
> I got BlockClosure>>doesNotUnderstand: #newProcessWithBindings:
>
> I searched a Seaside One click experience image (which is made with Pharo)
> and I found this method in *DynamicBindings category of BlockClosure. In
> order to make it work I simply filed out the whole category and loaded it in
> squeak. It seems to be working but I wanted to make sure I was doing it
> right. The thing is I want to show seaside to a friend that doesn't know
> anything about smalltalk so I don't want him to get frustrated with obscure
> bugs because of a bad installation. I suppose it would be best if I just
> give him the Seaside one click experience image but I really don't like
> Pharo.
> Also, I would like to know what other package do you think I should install
> in order to make a good starting environment for him.
> Thanks in advance.
>

[squeak-dev] Re: Create option in debugger window?

2009-10-07T22:04:27Z

Ricardo Moran wrote:
> I'm glad it's fixed now. That was fast!

It was an easy fix since it was just an oversight on my part. It was
just that the button wasn't hooked into the notifier properly.

Thanks again for reporting it. Can't fix stuff you don't know is broken ;-)

Cheers,
- Andreas

> On Thu, Oct 8, 2009 at 12:49 AM, Andreas Raab <andreas.raab@...
> <mailto:andreas.raab@...>> wrote:
>
> Hi Ricardo -
>
> That was just a bit of temporary breakage after the last ToolBuilder
> round. Thanks for pointing the problem out!
>
> Cheers,
> - Andreas
>
> Ricardo Moran wrote:
>
> Hi guys, first of all I want to say it's really nice to see
> Squeak finally moving forward. Yesterday I started working with
> the latest trunk image and I noticed that when a
> doesNotUnderstand error pops up the debugger window doesn't have
> the "create" button anymore. Maybe it was removed a lot of time
> ago, I don't know because I mostly work in Etoys.
> I really like that feature and I don't know how to put it back.
> Does anybody know how to do it?
>
> Cheers
>
>
> ------------------------------------------------------------------------
>
>
>
>
>
>
> ------------------------------------------------------------------------
>
>

Re: [squeak-dev] Re: Create option in debugger window?

2009-10-07T21:57:46Z

I'm glad it's fixed now. That was fast!

Cheers

On Thu, Oct 8, 2009 at 12:49 AM, Andreas Raab <andreas.raab@...> wrote:

Hi Ricardo -

That was just a bit of temporary breakage after the last ToolBuilder round. Thanks for pointing the problem out!

Cheers,
- Andreas

Ricardo Moran wrote:

Hi guys, first of all I want to say it's really nice to see Squeak finally moving forward. Yesterday I started working with the latest trunk image and I noticed that when a doesNotUnderstand error pops up the debugger window doesn't have the "create" button anymore. Maybe it was removed a lot of time ago, I don't know because I mostly work in Etoys.
I really like that feature and I don't know how to put it back. Does anybody know how to do it?

Cheers

------------------------------------------------------------------------

[squeak-dev] Re: Selection in new file list

2009-10-07T21:53:59Z

John M McIntosh wrote:

> Well we used URI with great success in Sophie, since we could use a
> relative path then resolve it to a network path, file system path, or a
> specialized in memory path in order to store book objects in those
> three storage areas.
>
> But a few things come to mind for example we could someone do
>
> foo := CPlatform homeDirectory or documentDirectory or tempDirectory
> or musicDirectory and it would mostly do the right thing based on the
> platform to construct a URI.

Interesting. The result was a URI instead of a FileDirectory? In
Qwaq^H^H^H^H Teleplace we we use e.g., FileDirectory
[temp|home|log|cache]Directory to return instances of FileDirectory. Or
did you implement most of the interesting file creation methods on URIs
as well?

> then what if you did
>
> fum := CPlatform homeDirectory , 'myStuff' , 'calledWhat' ,
> 'where.txt' .

Ooooh. Comma. I like that. Already a generic concatenation operator.
Much better than slash.

> using the ',' operator to concatenate a path together. Then a method to
> return the proper path based on the type you want, and of course the
> platform specific details.
>
> platformDirectoryPath = fum asFileSytemDirectoryPath.

That is quite nice. Yeah I could see using that. Any chance you can
re-release the URI code as MIT? (IIRC, Sophie is BSD, right?).

Cheers,
- Andreas

Re: [squeak-dev] Re: Selection in new file list

2009-10-07T21:44:02Z

Well we used URI with great success in Sophie, since we could use a
relative path then resolve it to a network path, file system path, or
a specialized in memory path in order to store book objects in those
three storage areas.

But a few things come to mind for example we could someone do

foo := CPlatform homeDirectory or documentDirectory or tempDirectory
or musicDirectory and it would mostly do the right thing based on the
platform to construct a URI.

then what if you did

fum := CPlatform homeDirectory , 'myStuff' , 'calledWhat' ,
'where.txt' .

using the ',' operator to concatenate a path together. Then a method
to return the proper path based on the type you want, and of course
the platform specific details.

platformDirectoryPath = fum asFileSytemDirectoryPath.

On 2009-10-07, at 9:04 PM, Andreas Raab wrote:

> K. K. Subramaniam wrote:
>> On Wednesday 07 Oct 2009 10:26:48 pm Nicolas Cellier wrote:
>>> In st80, there was also a nice feature: * and ? would also apply
>>> to any
>>> part of the hierarchy
>>> like: /home/nice/squeak/*/*.cs
>> Since Squeak is a VM accessing filesystem on its host, shouldn't we
>> be using URLs in the image and let FilePlugin resolve it into
>> pathnames?
>
> I'm not particularly in favor of that option in tools that actually
> expose the file names to users. When you type file paths in, it just
> feels wrong to type, e.g., /c/users/andreas/; and typing file:///C:/users/andreas/
> is just plain awful (yes, that's *three* slashes after the colon -
> copied right out of firefox).
>
> It would be more useful if slashes could be universally used in code
> though; having to write stuff like
>
> foobar := (FileDirectoy default directoryNamed: 'foo')
> directoryNamed: 'bar'.
>
> is every bit as awful. Perhaps a binary operator #/ would be in
> order? I.e.,
>
> foobar := FileDirectory default / 'foo' / 'bar'.
>
> (although this overloads the meaning of #/ which is not great either)
>
> Cheers,
> - Andreas
>

--
=
=
=
========================================================================
John M. McIntosh <johnmci@...> Twitter:
squeaker68882
Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com
=
=
=
========================================================================

[squeak-dev] Re: Selection in new file list

2009-10-07T21:04:41Z

K. K. Subramaniam wrote:
> On Wednesday 07 Oct 2009 10:26:48 pm Nicolas Cellier wrote:
>> In st80, there was also a nice feature: * and ? would also apply to any
>> part of the hierarchy
>> like: /home/nice/squeak/*/*.cs
> Since Squeak is a VM accessing filesystem on its host, shouldn't we be using
> URLs in the image and let FilePlugin resolve it into pathnames?

I'm not particularly in favor of that option in tools that actually
expose the file names to users. When you type file paths in, it just
feels wrong to type, e.g., /c/users/andreas/; and typing
file:///C:/users/andreas/ is just plain awful (yes, that's *three*
slashes after the colon - copied right out of firefox).

It would be more useful if slashes could be universally used in code
though; having to write stuff like

foobar := (FileDirectoy default directoryNamed: 'foo')
directoryNamed: 'bar'.

is every bit as awful. Perhaps a binary operator #/ would be in order? I.e.,

foobar := FileDirectory default / 'foo' / 'bar'.

(although this overloads the meaning of #/ which is not great either)

Cheers,
- Andreas

[squeak-dev] Re: Create option in debugger window?

2009-10-07T20:49:58Z

Hi Ricardo -

That was just a bit of temporary breakage after the last ToolBuilder
round. Thanks for pointing the problem out!

Cheers,
- Andreas

Ricardo Moran wrote:

> Hi guys, first of all I want to say it's really nice to see
> Squeak finally moving forward. Yesterday I started working with the
> latest trunk image and I noticed that when a doesNotUnderstand error
> pops up the debugger window doesn't have the "create" button anymore.
> Maybe it was removed a lot of time ago, I don't know because I mostly
> work in Etoys.
> I really like that feature and I don't know how to put it back. Does
> anybody know how to do it?
>
> Cheers
>
>
> ------------------------------------------------------------------------
>
>

[squeak-dev] The Trunk: Tools-ar.133.mcz

2009-10-07T20:46:26Z

Andreas Raab uploaded a new version of Tools to project The Trunk:
http://source.squeak.org/trunk/Tools-ar.133.mcz

==================== Summary ====================

Name: Tools-ar.133
Author: ar
Time: 7 October 2009, 8:46:02 am
UUID: 2f3b592e-b7d5-ba43-8f09-71b983afecd9
Ancestors: Tools-MAD.132

Put the 'create' button back into debug notifiers when we're encountering a DNU.

=============== Diff against Tools-MAD.132 ===============

Item was changed:
----- Method: Debugger>>buildNotifierWith:label:message: (in category 'toolbuilder') -----
buildNotifierWith: builder label: label message: messageString
+ | windowSpec listSpec textSpec panelSpec buttonSpec quads |
- | windowSpec listSpec textSpec panelSpec buttonSpec |
windowSpec := builder pluggableWindowSpec new.
windowSpec model: self.
windowSpec extent: 450 @ 156. "nice and wide to show plenty of the error msg"
windowSpec label: label.
windowSpec children: OrderedCollection new.

panelSpec := builder pluggablePanelSpec new.
panelSpec children: OrderedCollection new.
+ quads := self preDebugButtonQuads.
+ (self interruptedContext selector == #doesNotUnderstand:) ifTrue: [
+ quads := quads copyWith:
+ { 'Create'. #createMethod. #magenta. 'create the missing method' }
+ ].
+ quads do:[:spec|
- self preDebugButtonQuads do:[:spec|
buttonSpec := builder pluggableButtonSpec new.
buttonSpec model: self.
buttonSpec label: spec first.
buttonSpec action: spec second.
buttonSpec help: spec fourth.
panelSpec children add: buttonSpec.
].
panelSpec layout: #horizontal. "buttons"
panelSpec frame: (0@0 corner: 1@...).
windowSpec children add: panelSpec.

Preferences eToyFriendly | messageString notNil ifFalse:[
listSpec := builder pluggableListSpec new.
listSpec
model: self;
list: #contextStackList;
getIndex: #contextStackIndex;
setIndex: #debugAt:;
frame: (0@... corner: 1@1).
windowSpec children add: listSpec.
] ifTrue:[
message := messageString.
textSpec := builder pluggableTextSpec new.
textSpec
model: self;
getText: #preDebugMessageString;
setText: nil;
selection: nil;
menu: #debugProceedMenu:;
frame: (0@...: 1@1).
windowSpec children add: textSpec.
].

^windowSpec!

Re: [squeak-dev] Re: Installing seaside in squeak trunk image

2009-10-07T20:35:56Z

Ok, I'll give it a try. Thank you Andreas!

On Thu, Oct 8, 2009 at 12:34 AM, Andreas Raab <andreas.raab@...> wrote:

Hi Ricardo -

I have had better success with the up and coming 3.0 so you might give that a shot. In order to install Seaside3 into a trunk image do this:

"Update all the way from the trunk"
Utilities updateFromServer.

"Install the shiny new OB 2.0.3 release"
Installer installUrl:
'http://www.wiresong.ca/static/releases/OmniBrowser-2.0.3.sar'.

"Install Seaside 3.0"
(Installer squeaksource project: 'Seaside30')
install: 'LoadOrderTest'.

I have done this successfully as recently as a couple of weeks ago. If you find anything that doesn't work, let us know.

Cheers,
- Andreas

Ricardo Moran wrote:

Hi, I just installed Seaside version 2.8.4.591 from Universe. It finished ok but when I tried to start the server with
WAKomEncoded startOn: 8080.

I got BlockClosure>>doesNotUnderstand: #newProcessWithBindings:

I searched a Seaside One click experience image (which is made with Pharo) and I found this method in *DynamicBindings category of BlockClosure. In order to make it work I simply filed out the whole category and loaded it in squeak. It seems to be working but I wanted to make sure I was doing it right. The thing is I want to show seaside to a friend that doesn't know anything about smalltalk so I don't want him to get frustrated with obscure bugs because of a bad installation. I suppose it would be best if I just give him the Seaside one click experience image but I really don't like Pharo.
Also, I would like to know what other package do you think I should install in order to make a good starting environment for him.
Thanks in advance.

------------------------------------------------------------------------

[squeak-dev] Re: Installing seaside in squeak trunk image

2009-10-07T20:34:12Z

Hi Ricardo -

I have had better success with the up and coming 3.0 so you might give
that a shot. In order to install Seaside3 into a trunk image do this:

"Update all the way from the trunk"
Utilities updateFromServer.

"Install the shiny new OB 2.0.3 release"
Installer installUrl:
'http://www.wiresong.ca/static/releases/OmniBrowser-2.0.3.sar'.

"Install Seaside 3.0"
(Installer squeaksource project: 'Seaside30')
install: 'LoadOrderTest'.

I have done this successfully as recently as a couple of weeks ago. If
you find anything that doesn't work, let us know.

Cheers,
- Andreas

Ricardo Moran wrote:

> Hi, I just installed Seaside version 2.8.4.591 from Universe. It
> finished ok but when I tried to start the server with
>
> WAKomEncoded startOn: 8080.
>
> I got BlockClosure>>doesNotUnderstand: #newProcessWithBindings:
>
> I searched a Seaside One click experience image (which is made with
> Pharo) and I found this method in *DynamicBindings category of
> BlockClosure. In order to make it work I simply filed out the whole
> category and loaded it in squeak. It seems to be working but I wanted to
> make sure I was doing it right. The thing is I want to show seaside to a
> friend that doesn't know anything about smalltalk so I don't want him to
> get frustrated with obscure bugs because of a bad installation. I
> suppose it would be best if I just give him the Seaside one click
> experience image but I really don't like Pharo.
> Also, I would like to know what other package do you think I should
> install in order to make a good starting environment for him.
> Thanks in advance.
>
>
> ------------------------------------------------------------------------
>
>

[squeak-dev] Installing seaside in squeak trunk image

2009-10-07T20:01:25Z

Hi, I just installed Seaside version 2.8.4.591 from Universe. It finished ok but when I tried to start the server with

WAKomEncoded startOn: 8080.

I got BlockClosure>>doesNotUnderstand: #newProcessWithBindings:

I searched a Seaside One click experience image (which is made with Pharo) and I found this method in *DynamicBindings category of BlockClosure. In order to make it work I simply filed out the whole category and loaded it in squeak. It seems to be working but I wanted to make sure I was doing it right. The thing is I want to show seaside to a friend that doesn't know anything about smalltalk so I don't want him to get frustrated with obscure bugs because of a bad installation. I suppose it would be best if I just give him the Seaside one click experience image but I really don't like Pharo.

Also, I would like to know what other package do you think I should install in order to make a good starting environment for him.

Thanks in advance.

Re: [squeak-dev] Create option in debugger window?

2009-10-07T13:29:05Z

+1 i'm missing that too.
I often writing the code which sends not-yet implemented messages, and
then implementing them
on the fly by running the test code.

2009/10/7 Ricardo Moran <richi.moran@...>:

> Hi guys, first of all I want to say it's really nice to see
> Squeak finally moving forward. Yesterday I started working with the latest
> trunk image and I noticed that when a doesNotUnderstand error pops up the
> debugger window doesn't have the "create" button anymore. Maybe it was
> removed a lot of time ago, I don't know because I mostly work in Etoys.
> I really like that feature and I don't know how to put it back. Does anybody
> know how to do it?
> Cheers
>
>
>

--
Best regards,
Igor Stasenko AKA sig.

[squeak-dev] Box-Admins Team Report for August and September, 2009

2009-10-07T12:40:44Z

The box-admins team is responsible for seeing that the community shared
server is maintained and setting up any new features to be hosted on
that server.

For August and September, 2009:

I must apologize, I did not realize until today that I failed to send a
report for last month so consider this a cumulative report for the last
two months.

Not that there is all that much to say. The server has now been running
for 350 days and this team has only had very minor issues and questions
to deal with in the last 60 days.

http://www.squeak.org/Foundation/SupportSqueak/

and purchasing a Squeak badge. Note that a donation here may be used for
other Squeak community services including but not limited to
development, development infrastructure, and promotion of Squeak.

The services being hosted have continued to grow and include (but are
not limited to):

http://www.squeak.org/

SqueakMap - http://map.squeak.org/

http://ftp.squeak.org/

http://lists.squeakfoundation.org/

http://source.squeakfoundation.org/

http://map.squeak.org/

https://svn.squeak.org/

http://update.squeakfoundation.org/

http://wiki.squeak.org/

http://bugs.squeak.org/

and DNS services for squeak.org and squeakfoundation.org domain names.

Ken

signature.asc (197 bytes) Download Attachment

[squeak-dev] The Trunk: Kernel-nice.263.mcz

2009-10-07T11:58:27Z

Nicolas Cellier uploaded a new version of Kernel to project The Trunk:
http://source.squeak.org/trunk/Kernel-nice.263.mcz

==================== Summary ====================

Name: Kernel-nice.263
Author: nice
Time: 7 October 2009, 8:58:27 am
UUID: 2ca1dd7b-3aaf-ef40-a65d-9470d31bf6b6
Ancestors: Kernel-nice.262

Modify this:

(Object>>#printOn:) printString
before -> 'a CompiledMethod(1947: Object>>printOn:)'
after -> '(Object>>#printOn: "a CompiledMethod(1947)")'

This way, a CompiledMethod printString can be interpreted back.

This follows http://bugs.squeak.org/view.php?id=6726
A variant has been implemented in trunk previously, but in a non interpretable manner.
This one close the report.

=============== Diff against Kernel-nice.262 ===============

Item was changed:
----- Method: CompiledMethod>>printOn: (in category 'printing') -----
printOn: aStream
"Overrides method inherited from the byte arrayed collection."

+ aStream nextPut: $(; print: self methodClass; nextPutAll: '>>'; nextPutAll: self selector storeString; space; nextPut: $".
+ self printNameOn: aStream.
+ aStream nextPut: $(; print: self identityHash; nextPut: $); nextPut: $"; nextPut: $)!
- self printNameOn: aStream.
- aStream nextPut: $(; print: self identityHash; nextPutAll: ': ';
- print: self methodClass; nextPutAll: '>>'; nextPutAll: self selector; nextPut: $).
- "aStream space; nextPutAll: self identityHashPrintString"
- !

Re: [squeak-dev] Re: Selection in new file list

2009-10-07T11:42:43Z

On Wednesday 07 Oct 2009 10:26:48 pm Nicolas Cellier wrote:
> In st80, there was also a nice feature: * and ? would also apply to any
> part of the hierarchy
> like: /home/nice/squeak/*/*.cs
Since Squeak is a VM accessing filesystem on its host, shouldn't we be using
URLs in the image and let FilePlugin resolve it into pathnames?

Subbu

[squeak-dev] Create option in debugger window?

2009-10-07T10:17:12Z

Hi guys, first of all I want to say it's really nice to see Squeak finally moving forward. Yesterday I started working with the latest trunk image and I noticed that when a doesNotUnderstand error pops up the debugger window doesn't have the "create" button anymore. Maybe it was removed a lot of time ago, I don't know because I mostly work in Etoys.

I really like that feature and I don't know how to put it back. Does anybody know how to do it?

Cheers

Re: [squeak-dev] Re: Selection in new file list

2009-10-07T09:56:48Z

In st80, there was also a nice feature: * and ? would also apply to any part of the hierarchy
like: /home/nice/squeak/*/*.cs

Nicolas

2009/10/7 Bert Freudenberg <bert@...>

On 07.10.2009, at 17:41, Andreas Raab wrote:

> What do others think.

Two thoughts: First, I would only consider the last portion a pattern if it includes a wild card (there is little point in having a wild card that matches at most one file I think).

IMHO if there is neither a directory given nor a wild card, a substring search should be performed.

- Bert -

Secondly, I would walk the tree and select the directory "closest to" the chosen input. Meaning that if there's a typo you end up as close as we can make it to the chosen location.

Cheers,
- Andreas

Torsten Bergmann wrote:

The new File list is nice, but it still has a problem
with the selection in the tree after typing.
On win32 if one types C:\temp and hits ENTER the folder is not selected in the tree. If one enters C:\temp\ and ENTER it is selected. I think both options should be possible (since this is similar behavior as in Windows Explorer, and yes I'm biased). The reason is that anything after the "pathNameDelimiter"
is always identified as pattern.
See implementation in FileList>>pathAndPattern:
One possible solution could to check first if a subdirectory
with the extracted pattern exists
"(FileDirectory on: base) fileOrDirectoryExists: pat"
and react accordingly (ignore it as pattern and add it to the
base).
What do others think. Bye
T.

Re: [squeak-dev] Re: Selection in new file list

2009-10-07T08:51:19Z

On 07.10.2009, at 17:41, Andreas Raab wrote:

> > What do others think.
>
> Two thoughts: First, I would only consider the last portion a
> pattern if it includes a wild card (there is little point in having
> a wild card that matches at most one file I think).

IMHO if there is neither a directory given nor a wild card, a
substring search should be performed.

- Bert -

> Secondly, I would walk the tree and select the directory "closest
> to" the chosen input. Meaning that if there's a typo you end up as
> close as we can make it to the chosen location.
>
> Cheers,
> - Andreas
>
> Torsten Bergmann wrote:
>> The new File list is nice, but it still has a problem
>> with the selection in the tree after typing.
>> On win32 if one types C:\temp and hits ENTER the folder is not
>> selected in the tree. If one enters C:\temp\ and ENTER it is
>> selected. I think both options should be possible (since this is
>> similar behavior as in Windows Explorer, and yes I'm biased). The
>> reason is that anything after the "pathNameDelimiter"
>> is always identified as pattern.
>> See implementation in FileList>>pathAndPattern:
>> One possible solution could to check first if a subdirectory
>> with the extracted pattern exists
>> "(FileDirectory on: base) fileOrDirectoryExists: pat"
>> and react accordingly (ignore it as pattern and add it to the
>> base).
>> What do others think. Bye
>> T.
>
>

[squeak-dev] Re: Selection in new file list

2009-10-07T08:41:37Z

> What do others think.

Two thoughts: First, I would only consider the last portion a pattern if
it includes a wild card (there is little point in having a wild card
that matches at most one file I think).

Secondly, I would walk the tree and select the directory "closest to"
the chosen input. Meaning that if there's a typo you end up as close as
we can make it to the chosen location.

Cheers,
- Andreas

Torsten Bergmann wrote:

> The new File list is nice, but it still has a problem
> with the selection in the tree after typing.
>
> On win32 if one types C:\temp and hits ENTER the folder
> is not selected in the tree.
>
> If one enters C:\temp\ and ENTER it is selected. I think
> both options should be possible (since this is similar
> behavior as in Windows Explorer, and yes I'm biased).
>
> The reason is that anything after the "pathNameDelimiter"
> is always identified as pattern.
> See implementation in FileList>>pathAndPattern:
>
> One possible solution could to check first if a subdirectory
> with the extracted pattern exists
> "(FileDirectory on: base) fileOrDirectoryExists: pat"
> and react accordingly (ignore it as pattern and add it to the
> base).
>
> What do others think.
>
> Bye
> T.