David,

I can actually commit the patch #2 without problem; I'm just worried about the side effects (if any) that this could cause on the existing code. For example, I don't know if the set-service-fields operation is copying also the user login from one map to another one; if this is happening, then (with the existing code) the user login is overriden by the one of the calling service; if I apply my patch this will be no more true... could this cause problems in your opinion? Or am I just too anxious?

Thanks for your feedback,

Jacopo



David E. Jones wrote:

Sorry for the delay on this Jacopo.

Looking at this I actually like the fix in #2, ie changing the simple-method operations because what they do doesn't make sense. In other words, if you add a userLogin yourself you wouldn't expect the system to blow it away.

If you agree feel free to commit that, your patch of the CallService and CallServiceAsynch classes looks just fine. If I don't see your response on it I'll throw that in later this evening or tomorrow.

-David


On Feb 7, 2007, at 10:38 AM, Jacopo Cappellato wrote:

Ah,

forgot to mention that I'd vote for solution #1.

Jacopo

Jacopo Cappellato wrote:
Should we take some action about this subject?
I know that probably my explanations were not clear enough, so I'm proposing the two ways of fixing what I think it's an outstanding issue; I hope that, reviewing the (extremely simple) fixes I'm proposing will be easier to understand the issue I'm trying to report. Solution #1: just fix the service call to "createUserLogin" inside of createPersonAndUserLogin by adding the include-user-login="false"; in this way the manually "system" user, manually put in the input context will be not overriden by the user available in the context of the calling service ("createPersonAndUserLogin"); see fix1.patch Solution #2: fix for the minilang service call methods; if include-user-login is not set or it is true, the user login is put in the context but *only* if in the context there is not already a user login; see fix2.patch
I really think that one of the two solutions should be committed.
What do you think?
Jacopo
Jacopo Cappellato wrote:
David,

I really did not describe it very well, sorry.
I don't think there is a problem with the createUserLogin service definition or implementation. The issue (maybe) could be the way the framework handles the minilang attribute include-user-login="true": it just gets the "userLogin" object from the context of the calling service and put it in the context of the called service, even if in the context of the called service there is already a "userLogin" object. So for example the code in the createPersonAndUserLogin service (for which I have provided the patch) is:

<!-- call the service with the system account to get around security constraints for this special create --> <entity-one entity-name="UserLogin" value-name="createUlInMap.userLogin">
    <field-map field-name="userLoginId" value="system"/>
</entity-one>

<call-service service-name="createUserLogin" in-map-name="createUlInMap" include-user-login="false"/>

Unfotunately the "system" userLogin object that is manually put in the input context for the createUserLogin service is overriden (by the framework), if one is available in the context of the createPersonAndUserLogin service because the include-user-login attribute defaults to true.

Just adding the include-user-login="false" attribute to the service call fixed the issue.

Does it make sense?

Jacopo



David E. Jones wrote:

On Jan 25, 2007, at 7:41 AM, Jacopo Cappellato wrote:

Hi all,

could you have a look at the attached simple patch? It fixes an authorization problem under some special situations (switching from anonymous user to authenticated one...); however the issue is that, when the "createUserLogin" is called, if the attribute include-user-login is not set to false the manually passed in "system" user is overwritten by the user in the context.

I just traced through the code to try to figure out why this might be happening, and either I don't understand what you're describing, or I just can't find why this would happen.

In the implementation of the service (LoginServices.java:createUserLogin) I don't see anything that would do this. In the service definition I don't see it either because the service definition does not have a "userLogin" parameter going out.

Have you traced down the code that would be putting the user in the context? Also, do you mean by this the context of the current service call (ie for other services called by ECA rules, etc), or in the result (returned) Map of the service, or somewhere else?

I apologize for pushing on the detail, I may just be too brain-dead to see it right now.

-David


Should I commit this patch?
Or, in general, would be better, even if include-user-login is true, to set the user login only if one is not already there in the service in map?

Jacopo
Index: applications/party/script/org/ofbiz/party/party/PartySimpleMethods.xml
===================================================================
--- applications/party/script/org/ofbiz/party/party/PartySimpleMethods.xml (revisione 499802) +++ applications/party/script/org/ofbiz/party/party/PartySimpleMethods.xml (copia locale)
@@ -85,7 +85,7 @@
             <field-map field-name="userLoginId" value="system"/>
         </entity-one>

- <call-service service-name="createUserLogin" in-map-name="createUlInMap"/> + <call-service service-name="createUserLogin" in-map-name="createUlInMap" include-user-login="false"/> <entity-one entity-name="UserLogin" value-name="newUserLogin"/>

         <field-to-result field-name="newUserLogin"/>


------------------------------------------------------------------------
Index: applications/party/script/org/ofbiz/party/party/PartySimpleMethods.xml
===================================================================
--- applications/party/script/org/ofbiz/party/party/PartySimpleMethods.xml (revisione 503593) +++ applications/party/script/org/ofbiz/party/party/PartySimpleMethods.xml (copia locale)
@@ -85,7 +85,7 @@
             <field-map field-name="userLoginId" value="system"/>
         </entity-one>
- <call-service service-name="createUserLogin" in-map-name="createUlInMap"/> + <call-service service-name="createUserLogin" in-map-name="createUlInMap" include-user-login="false"/>
         <entity-one entity-name="UserLogin" value-name="newUserLogin"/>
          <field-to-result field-name="newUserLogin"/>
------------------------------------------------------------------------
Index: framework/minilang/src/org/ofbiz/minilang/method/callops/CallServiceAsynch.java
===================================================================
--- framework/minilang/src/org/ofbiz/minilang/method/callops/CallServiceAsynch.java (revisione 503593) +++ framework/minilang/src/org/ofbiz/minilang/method/callops/CallServiceAsynch.java (copia locale)
@@ -68,8 +68,9 @@
         if (includeUserLogin) {
             GenericValue userLogin = methodContext.getUserLogin();
 -            if (userLogin != null)
+            if (userLogin != null && inMap.get("userLogin") == null) {
                 inMap.put("userLogin", userLogin);
+            }
         }
                  // always add Locale to context unless null
Index: framework/minilang/src/org/ofbiz/minilang/method/callops/CallService.java
===================================================================
--- framework/minilang/src/org/ofbiz/minilang/method/callops/CallService.java (revisione 503593) +++ framework/minilang/src/org/ofbiz/minilang/method/callops/CallService.java (copia locale)
@@ -213,7 +213,7 @@
         if (includeUserLogin) {
             GenericValue userLogin = methodContext.getUserLogin();
 -            if (userLogin != null) {
+            if (userLogin != null && inMap.get("userLogin") == null) {
                 inMap.put("userLogin", userLogin);
             }
         }





Reply via email to