Seam's power is such that you're able to directly expose EntityHome objects in
the view and use pages.xml to wire it all together, and even provide security.
However, if you prefer to do things in code, just think of EntityHome as an
implementation of the "Generic DAO" pattern that plays nice w
This is not a good way to get people to write your program for you. Anyway,
seam-gen will generate almost all this for you. If you find seam-gen isn't
sufficient for the purpose, consider filing a feature request instead of some
challenge (which in my inbox looked a whole lot like spam)
Vi
I don't typically use a star for required fields, but a distinct background
behind the form field and label. Would it be possible to use
"(before|after)RequiredField" and "aroundRequiredField", the way it's done for
errors?
View the original post :
http://www.jboss.com/index.html?module=bb
Egads, nevermind ... I'd forgotten about the overloaded forms of
@Local/@Remote.
At any rate, you still need to ask on an EJB3 forum, or you'll get
misinformation like what you just got from me.
View the original post :
http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4026751#40267
You need to read the EJB documentation. The JBoss EJB3 trailblazers are quite
nice. @Local and @Remote belong on your interface, not your implementation.
View the original post :
http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4026746#4026746
Reply to the post :
http://www.jboss.co
Both EntityHome instances should use the same conversation-scoped EntityManager
(a component named "entityManager" by default), so they'll guaranteed
participate in the same JTA transaction. No need to do any extra
"synchronizing".
View the original post :
http://www.jboss.com/index.html?mo
If they really are unrelated, why not just use two EntityHome instances? If
they are actually related somehow, you can express the relation in
JPA/Hibernate terms as @OneToOne (or whatever your relation is) and get at one
object through the other -- no entityhome required.
@EntityHome seems t
You could make personHome inject the current phone, then have
personHome.persist() call getInstance().phones.add(phone)
However, that wires the policy a little too low level for my tastes. I
personally recommend an EditPersonAction (or whatever you want to call it) to
back your edit page that
Versioning and optimistic locking is also immune to the inevitable race
condition of checking for changes before a commit. It does throw an exception
though, which makes recovery like offering a "merge" screen somewhat
complicated.
Sometimes you still might want to peek outside the current tra
Propagation won't recreate the conversation, and your conversation is still
active if you didn't end it yourself. However, your entityManager is toast, so
you need a new one, and that's usually done with a new conversation. It should
be possible to recreate the entityManager within the current
Of course you're getting stale data -- you're using SESSION scope and an
extended persistence context that never changes during the bean's lifetime.
Same PC, same isolated view of the database for the whole session.
Seam certainly doesn't discourage using Stateful Session Beans, it discourages
"[EMAIL PROTECTED]" wrote : I've just committed a new feature to CVS that lets
code changes made to Seam JavaBean components be visible immediately without a
container restart.
Mmmm yeah, take that tapestry 5. I think I'm going to have to migrate away
from EJB's sooner rather than later to tak
Done, JBSEAM-1002.
I just got to thinking that it would be neat if ComponentScanner and
NamespaceScanner were themselves components that could be overridden. That way
it'd be simple for anyone to add their own app-specific validation logic. Does
make for a bit of a bootstrap problem, I'll adm
It's nice that there's more than one way to create and name components in seam,
but it certainly takes some discipline on the part of the programmer to
organize them thoughtfully, and minor slips in naming result in null values
getting injected. This is easy to catch when not using required=fal
A lot of this depends on your specific application: if the set of tasks is a
fixed "wizard" workflow, then you can probably get by with conversations alone
(using manual flush mode if it involves editing "live" entities). If your
tasks themselves are varying or defined by complex business logic
Why not just try it and see? And yes, I would recommend using Tomcat. It's
possible to slim down JBoss quite a bit, but it does have a good size out of
the box footprint in the profile that's required for Seam.
View the original post :
http://www.jboss.com/index.html?module=bb&op=viewtopic&p=
My version of "production" varies quite a bit from that of an e-commerce site,
and that itself is quite different than what a bank might expect.
At any rate, Seam performs about as well as your JSF and EJB implementations.
The overhead of Seam itself is pretty miniscule compared to the (large)
Okay, subtract me from the people using @LoggedIn -- it's simply evil, and it
seems to interfere with bijection and @Create methods. Until I devote some
time to reading up on Seam Security, I've literally hardcoded the login check
into every method that would use it (which is actually not many
I've been using @LoggedIn. It's a hack, but if it's good enough for the
booking example, it's good enough for my simple CRUD app. I tweaked it a bit
to return a "nologin" result in order to redirect to the login page when users
attempt an action that requires a login.
I'd love to know how to
I'm a dunderhead -- not that this is a big surprise to anyone. It had nothing
to do with transaction isolation levels, and everything to do with needing to
set flushMode=MANUAL when beginning the conversation. Once I did that,
everything works as expected, and entities aren't flushed until I a
I still can't explain the mysterious breakage that started when using another
entityManager, but it went away when I switched from a javabean to a SFSB,
something I was going to do eventually anyway. I was still seeing dirty data,
but I found it was the database I was using -- it looks like HSQ
Actually it looks like simply using another EntityManager anywhere in my action
bean caused a completely _different_ component (ruleTypes) in the action bean
to not be injected. Same result whether I injected it with @PersistenceContext
or created it by hand with new EntityManagerFactory().crea
I should probably clarify, the exception gets thrown the first time, but the
second update appeared to succeed when in fact it silently dropped. I've done
quite a bit of rearranging now, and haven't reproduced the problem.
As for checking the unique constraint by hand before possibly violating
In the framework, I'm trying to enforce optimistic locking and a unique
constraint on the "name" attribute. For optimistic locking, there's not a lot
I can do except kick them out (they can always open a new window and use their
back button to copy their changes and merge them in). For name co
The conversation timeout controls how quickly open conversations expire -- it's
the direct equivalent of session timeouts, but for conversations. It has
nothing to do with server timeouts, which are specific to the webserver. I
imagine for JBoss, it would be somewhere in the tomcat configurati
SFSBs and Entities are bound to the conversation context. SLSBs are, obviously,
in the stateless scope.
This is all covered in the Seam documentation.
http://docs.jboss.com/seam/latest/reference/en/html/index.html
View the original post :
http://www.jboss.com/index.html?module=bb&op=viewto
Calling WTP is a little bit flakey is like calling Ubuntu a little bit brown.
I've actually given up on WTP's crummy "publish" mechanism and gone and written
a shell script to jar up and deploy the ear by hand. And unlike WTP's
mouse-only (afaict) interface for manual publish, I can just make
We do that all the time ;)
Seriously, it does make it easier to google solutions for a problem when one
talks themself through the solution publicly.
chuck
View the original post :
http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3980168#3980168
Reply to the post :
http://www.jboss.c
I wasn't aware that it existed. The more control of UserTransaction that's
possible, the better -- though I imagine anyone using UserTransaction is
probably already controlling it through code and not declaratively, so
injection would probably suffice.
View the original post :
http://www.jb
mpleteness. Is this omission
intentional?
chuck (still no luck getting my login recognized as chuckadams and not cja987)
View the original post :
http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3980160#3980160
Reply to the post :
http://www.jboss.com/index.html?module=bb&
I added the "messages" bundle to the bundle names list, and copied an existing
messages.properties file from the examples. It's a workaround I can deal with
for now, but not one I'm comfortable with -- I doubt I have all the messages
available should I add new validators.
I'm sort of puzzled
I'm working with my optimistic-locking contact manager example, and added a
messages.properties to the root of my source file with just one message in it:
| staleEditContact=Sorry, someone else has already updated this contact while
you were working on it.
|
And now I get this as soon as
I was going by the description of page actions in the seam docs, section 3.1.1
anonymous wrote :
| The page action method can return a JSF outcome. If the outcome is
non-null, Seam will delegate to the defined JSF navigation rules and a
different view may end up being rendered.
|
I took t
Aha, that did it, and I get ... yay, an HTTP 500 .. but the right error on the
console. I guess dealing gracefully with the exception means catching it in
update() and company. Figuring out how to present a UI to merge changes should
be an interesting project.
I can't see anything in @Begin's
Here you go:
| 16:32:06,726 DEBUG [Naming] JNDI InitialContext
properties:{java.naming.factory.initial=org.jnp.interfaces.NamingContextFactory,
java.naming.factory.url.pkgs=org.jboss.naming:org.jnp.interfaces}
| 16:32:06,726 DEBUG [AbstractSeamPhaseListener] beginning transaction prior
to
It took much fighting with log4j.xml, but I got a debug log. The whole log is
1340 or so lines long, so I'll see what I can glean from it
When I click "Edit Contact", I do see it creating a conversation:
| 16:32:06,766 DEBUG [Manager] No stored conversation, or concurrent call to
the stored
my log4j kung fu is poor, so I just added this to server/default/conf/log4j.xml
and restarted, to no avail.
|
|
|
|
I have tried many other things as well, but I remain defeated. How do I enable
debug logging?
View the original post :
http://www.jboss.com/index.html
"[EMAIL PROTECTED]" wrote : I bet you don't have the SeamRedirectFilter
installed.
It's definitely installed.
|
| Seam Redirect Filter
|
| org.jboss.seam.servlet.SeamRedirectFilter
|
|
|
|
|
I had thought there would be some funniness with a conversation-scoped
component starting its own conversation, so I moved beginEdit() to its own
class (better design anyway) and edited pages.xml accordingly
| @Name("contactAction")
| public class ContactAction {
|
| @Logger
|
I'm probably doing this completely wrong... I took contactHome out of
components.xml and replaced it with this class:
| @Name("contactHome")
| @Scope(ScopeType.CONVERSATION)
| public class ContactHome extends EntityHome {
|
| @Logger
| private Log log;
|
| publi
The "war" target in seam's build.xml used by the examples does some filterset
tweaking for the jndiPattern and to turn the embedded ejb off, and so on. So
while your sources are the same, if you use a different build process then you
won't get the same results. Just tweak your components.prope
Newbie question alert... I'm working with the contactlist sample app (works
fine now that I'm back to using myfaces) and I want to prevent lost updates. I
added a version field to Contact, gave it a @Version annotation, and a
getter/setter pair, and it seems to get updated by Hibernate as expec
PEBKAC again, I forgot to update the views themselves.
Still, the problems aren't over: when I edit a contact and click update, it
switches immediately to the create screen and offers me "Create Contact"
instead (which does actually work). Deleting a contact blanks all the fields,
but doesn't
Okay, I switched all the .xhtml extensions in pages.xml to .seam, and now
viewContact works, but editContact continues to give me a blank contact.
Delete Contact is also nonfunctional, and results in this on the console:
WARNING: phase(RENDER_RESPONSE 6,[EMAIL PROTECTED]) threw exception:
java
Heh, all that typing and I could have waited a few more minutes instead. I'll
try using the .seam view ids and see how that works.
View the original post :
http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3979189#3979189
Reply to the post :
http://www.jboss.com/index.html?module=b
Positive. In fact I forgot to set it back to true when I tried it with
myFaces, and it still worked.
If the contactlist example works out of the box with the 1.2RI, then perhaps
I'm configuring it wrong in jboss. Here's what I did to run it:
* cvs checkout and build seam
* delete myfaces*
I can't get the contactlist example working with the 1.2 RI -- when I try to
edit or view a contact, I get a blank contact instead. I wonder whether some
of my other hear-tearing experiences getting CRUD working by hand is related to
the same problem.
I opened up JBSEAM-427 in JIRA on this, bu
I am also having the same problem when using the RI with Seam CVS. This
occurred with my own app and with the booking example.
View the original post :
http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3976653#3976653
Reply to the post :
http://www.jboss.com/index.html?module=bb&op=po
This sounds to me a lot like the "Flash" scope of RoR and family. Though
conversations don't look very heavyweight to me, a flash scope could be nice in
order to remove any need to use that "conversationId" (or cid as it is in cvs)
parameter.
That segues to a minor gripe of mine, namely the co
Many of the examples are using bare CR's for line endings, which makes many
editors (such as vi and emacs) quite unhappy. I imagine they're checked into
CVS as binary, when they should be text. CVS will translate line endings
appropriately if they're checked in as text.
The list of files th
Just out of curiousity, does the JBoss installed with the NetBeans+JBoss bundle
have a configuration that's usable with Seam?
View the original post :
http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3975833#3975833
Reply to the post :
http://www.jboss.com/index.html?module=bb&op=post
51 matches
Mail list logo
