Korakurider wrote:
> p.s having alpha release image would make sense if updating is such
> troublesome !
Bert posted a couple of images that seem to meet this description at
http://etoys.squeak.org/download/
They are made by hand, and there is no guarantee of regular
replacements. (An automatic build host would have failure modes that
are similar to the update button.)
I came across a possible issue. I found a project where textual code
would not compile, because Etoys couldn't
resolve the name of a player to a class.
I opened this project of Bert's from 2008
http://wiki.squeak.org/squeak/uploads/Carousel.pr
with Etoys 4
There is a textual script pane already open. I made a small edit to
it (changed one of the constants) and typed command-s to accept it.
Squeak's pop-up menu for unrecognized names appeared: it didn't
recognize Carousel.
It seems the code itself was fine. I was able to discard the script,
and make a new one with the same name, and paste the code into it. I
could then edit and compile it to my heart's content, to modify the
animation.
I don't know enough about the compiler to fix it. If you think this
is an issue, I will post it in the bug tracker, and leave it to others
to verify and perhaps fix.
David
(I'm answering this on-list)
On 25.05.2010, at 01:35, Ricardo Moran wrote:
> I tried to update my image and I got an emergency evaluator. I was about to report the issue when I realized my image was the problem, other images updated just fine. Anyway, I managed to update correctly (I think) by changing the "GetText-Localization" category to "System-Localization". I'm not sure why that worked, though.
What happened is that this update:
http://etoys.squeak.org/updates/2375repo1117-bf.cs
loaded GetText-bf.2 before System-bf.16. This would be the right order if you had not had GetText-Richo.1 in your image yet (as when you update a clean image). But since I moved back stuff from GetText to System, loading GetText first actually removed a few essential classes from your image (they would have been added back when loading System, but it failed before).
Moving classes and methods between packages is delicate in Monticello. With a series of package configurations it can be made to work fine. E.g., compare how these configs differ in the order of GetText and System packages:
http://source.squeak.org/etoys/update-bf.3.mcmhttp://source.squeak.org/etoys/update-bf.4.mcm
In update #2375 I combined these two configs into one, which worked in a clean image. I now think that was a mistake - had I made a changeset with these two configs it should have worked fine even in your image. Lesson learned.
A better way would be to load those packages together rather than sequentially, but our current Monticello version does not support that. Also, the sequential loading at least is predictable ...
> Just to be sure, the latest version is 'etoys4.1 of 29 April 2010 update 2378 (repo v1117 of 24 May 2010)', right?
Yes.
And welcome as an Etoys developer - you have commit access to the main repo now! :)
- Bert -
On 23.05.2010, at 21:01, commits(a)source.squeak.org wrote:
>
> A new version of MorphicExtras was added to project Etoys Inbox:
> http://source.squeak.org/etoysinbox/MorphicExtras-Richo.2.mcz
There already is a MorphicExtras-Richo.3 version in the main repository, it would be rather confusing if I merged this ...
- Bert -
Ricardo Moran uploaded a new version of ReleaseBuilder to project Etoys:
http://source.squeak.org/etoys/ReleaseBuilder-Richo.2.mcz
==================== Summary ====================
Name: ReleaseBuilder-Richo.2
Author: Richo
Time: 25 May 2010, 12:18:41 am
UUID: 108b4a0a-249b-0144-8810-ee61c1298b55
Ancestors: ReleaseBuilder-bf.1
* Changed a reference to GetTextExporter2 to GetTextExporter
=============== Diff against ReleaseBuilder-bf.1 ===============
Item was changed:
----- Method: ReleaseBuilderSqueakland>>updateGettext (in category 'olpc') -----
updateGettext
"ReleaseBuilderSqueakland new updateGettext"
"Export gettext template and import all translation in po/"
+ GetTextExporter exportTemplate.
- GetTextExporter2 exportTemplate.
GetTextImporter importAll.!
Ricardo Moran uploaded a new version of MorphicExtras to project Etoys:
http://source.squeak.org/etoys/MorphicExtras-Richo.4.mcz
==================== Summary ====================
Name: MorphicExtras-Richo.4
Author: Richo
Time: 24 May 2010, 9:23:55 pm
UUID: f3747e9a-ae24-804d-bc3a-1298afada9bc
Ancestors: MorphicExtras-Richo.2, MorphicExtras-Richo.3
* The unregister flaps quads error appeared again.
(see http://lists.squeakland.org/pipermail/etoys-dev/2010-May/004658.html).
This is not the best fix but at least it lets you finish the update.
=============== Diff against MorphicExtras-Richo.3 ===============
Item was changed:
----- Method: Flaps class>>unregisterQuadsWithReceiver: (in category 'flaps registry') -----
unregisterQuadsWithReceiver: aReceiver
"delete all quads with receiver aReceiver."
self registeredFlapsQuads
do: [:assoc | assoc value
+ removeAllSuchThat: [:q | (self environment at: (q first) ifAbsent: [nil]) = aReceiver ]]!
- removeAllSuchThat: [:q | (self environment at: (q first)) = aReceiver ]]!
Hi,
I have tested the CameraMorph project and it is very nice.
You can get it here.
http://www.doconnel.force9.co.uk/squeak/CameraPlugin-Linux.tar.gz
I had to download
http://download.scratch.mit.edu/WinScratch1.4.zip
Unzip and get the CameraPlugin.dll and put it in the Etoys vm directory to
meke this work on Windows XP.
So far it looks good. Filed in files as described it ReadMe.
I did not find any bugs yet :-)
It took a few moments for me to find the settings to turn the camera on,and
track color etc.
Maybe these controls could be added to the CameraMorph handle menu as well ?
Maybe a row of controls on the CameraMorph it self ?
Camera:controll live feed, next frame and skip frames. It is a little
unclear what all these controls are for
Ah, found a bug:
Set Camera's frame rate to zero: BANG!
In Camera:tracking we should enable World geometry to be able to get the
World dimensions so 'Camera's world height' and 'Camera's world width' could
be removed...
If the tracking could be spun off in a primitive it would be possible to get
much better result
The FX where not part of this package.
Great work
Karl