Hi,
I'm wrestling with ideas for a prototyping platform (ultimately possibly
production platform) using Seaside and possibly Gemstone as an object
store. I'm new to Smalltalk so I'm wrestling with great number of concepts
to start with. I've recently been working through the Seaside book and the
section on Magritte look very interesting. I'm working with VisualWorks 7.7
and when I try to load the package of the Public store I get a number of
errors and it seems to do a partial load of the Magritte classes (I have
packages created and some methods injected in existing classes.)
My questions are the following.
1. Is Magritte actively maintained to a point where it's usable ?
2. Does anyone have experience with loading MagritteForVisualWorks from the
cincom public repository (I have turned off atomic loads)?
Best Regards,
Miguel
Folks,
I'm running Pier from the piercms.com one click image, and writing my
own structure and view subclasses, etc.
I'm wanting to produce an anchor which when clicked, will perform a
command and then switch to a different view.
I can do either one of those things, but seemingly not both. My
assumption was that I'd need to write something like the following,
where #view: is sent to the result of #command: ...
html anchor
goto: (
(self context command: MyCommandClass new) view: MyViewClass
);
with: 'click me'.
But I find that while this code takes me to the new view, the command is
never executed.
Is there an established way to do this? If not, I'd appreciate any ideas
on the best approach. It occurs to me that I could subclass the view
command and put my desired command behaviour in that.
Cheers,
Nick Brown
At 11:14 11/04/2010, Lukas Renggli wrote:
>normally Pharo cannot be updated in an image that has
>something else loaded but the core packages.
Good to know.
It took time, but at the end I was happy with the result 8-)
(although, have not yet tested it thoroughly)
thanks!
>There are always daily builds at <http://hudson.lukas-renggli.ch/>
>with the latest Pharo, Seaside, Magritte and Pier code. We will soon
>update the one-click image too (based on these builds).
That would be great, although, in general, I do prefer updating my
image manually, to get a chance to have a look at the new code during
the merge process.
Take care,
Reza
At 10:44 11/04/2010, Lukas Renggli wrote:
>How did you load the Seaside 3 code?
Just picked this one:
<http://www.piercms.com/download>http://www.piercms.com/download
Then upgraded to *PharoCore1.0-#10517*, and the
latest Seaside 3.0, and Pier 2 packages,
including FFI, RFB, OS-Process, File-Kernel, etc.
It took me several hours :-( If the result could
help somebody saving that time, please let me
know to put the image+changes (~50 M°, unzipped)
on a public server for download.
Cheers,
Reza
At 10:00 11/04/2010, Lukas Renggli wrote:
> > Using Monticello browser, I just updated a standard 1.2 Pier image to the
> > latest versions of Pharo (PharoCore1.0-#10517), Seaside 3 and Pier 2.
>
>Generally updating from Seaside 2.8 to Seaside 3.0 does not work, due
>to the lack of atomic loading of Monticello. Take a fresh image.
Hi Lukas,
Actually, I just loaded the latest packages into Seaside 3.0 + Pier 2
image. I don't use Seaside 2.8 and Pier 1 since a few months.
I just discovered that it was related to the cached
WARequestHandlingConfiguration. Just executed *WAEnvironment
reloadApplications*, and everything is in order now ;-)
Thanks for your support,
Regards,
Reza
Hi all,
Using Monticello browser, I just updated a standard 1.2 Pier image to
the latest versions of Pharo (PharoCore1.0-#10517), Seaside 3 and Pier 2.
During the update process, WAAttributeNotFound was raised
upon the call to #preferenceAt: #scriptGeneratorClass put: aClass
by the following methods:
- JQAllFunctionalTests class >>initialize
- JQUiAllFunctionalTestsclass >>initialize
- SUAllComponentTestsclass >>initialize
- SUAllFunctionalTestsclass >>initialize
I had to bypass the issue by temporarily the code of those methods.
Now, the same exception is raised again,
when trying to connect to Pier via the local host
(<http://localhost:8080/pier>http://localhost:8080/pier),
upon the call to *self handler configuration at:
#scriptGeneratorClass*,
in WARequestContext >> newDocument.
#newDocument has been changed on Feb, 22, 2010.
Does anybody see how this issue could be better fixed; just reverting
to the previous code doesn't seem a good idea.
Thanks in advance,
Reza
I tried to do some tests about transferring data via context from one page to another in pier. I did
renderContentOn: html
| ctx |
self context propertyAt: #forwardText put: 'set on old context'.
ctx := self context structure: (PRPathLookup start: self context structure path: '/targetpage').
ctx propertyAt: #forwardText put: 'set on new context'.
html anchor
goto: ctx;
with: 'go'
On the targetpage there is a component that reads the property from the context and displays it. What makes me wonder is that under any circumstances the text being display is "set on old context". Even if I press the link "go". Why is that so?
Norbert
Hi,
I have description that goes something like this:
Autor>>descriptionRegija
^ MASingleOptionDescription new
accessor: #regija;
label: 'Regija';
priority: 700;
options: LAURegija findAll;
beRequired;
yourself
the idea is to have a drop down that will display all available LAURegija
instances. Now the problem is that Magritte seems to cache this description
and method descriptionRegija does not get called every time the form is
displayed, and consequently, if a new LAURegija is added, it is not present
in the drop down.
So how to work around it?
Many thanks,
rush
http://www.cloud208.com/
Hi,
in gemstone stream next throws an exception if the stream is atEnd. In pharo it returns nil. I changed in
PBBlog>>defaultChildName
code next
ifNotNilDo: [ :separator | stream nextPut: separator ] ] ]
to
code atEnd
ifFalse: [ stream nextPut: code next ] ] ]
Norbert
I installed a complete new pier on a fresh gemstone image and noticed that the RSS views are not working. This is due another "aString is not equal to aSymbol" issue.
In PRPierFrame class>>classFromRequest: aRequest name: aString base: aBaseClass
there is
^ aBaseClass withAllConcreteClasses
detect: [ :each |
each isAbstract not
and: [ each name = value
or: [ each label = value ] ] ]
ifNone: [ nil ]
but class names are symbols and http parameters are strings. So I changed this to
^ aBaseClass withAllConcreteClasses
detect: [ :each |
each isAbstract not
and: [ each name = value asSymbol
or: [ each label = value ] ] ]
ifNone: [ nil ]
Norbert
