You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@sling.apache.org by "Krystian Nowak (Issue Comment Edited) (JIRA)" <ji...@apache.org> on 2012/01/26 16:58:38 UTC

[jira] [Issue Comment Edited] (SLING-2191) Allow multiple OptingServletS for the same resource type

    [ https://issues.apache.org/jira/browse/SLING-2191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13193882#comment-13193882 ] 

Krystian Nowak edited comment on SLING-2191 at 1/26/12 3:56 PM:
----------------------------------------------------------------

The problem described in this issue doesn't seem to be an edge case for me. Please look at my example this time:

We have two (and possibly more) implementations of the OptingServlet interface. They are configured to work on the same resourceType, the same method and no selectors. Each of them chooses to accept a request based on a request parameter. For us they should be able to work independently as long as the accept logic for is mutually exclusive for both of them.
Sadly this is not the case. Only one of them is chosen to handle the request and if the selected servlet decides to not accept it, the other servlet is omitted. The resolver seems to choose the handling servlet based on an order determined while registering (starting bundle) - we are using Felix here.

We could of course choose to use different resourceTypes or selectors but this does not solve the problem, as there are multiple applications on the platform and each one of them can potentially break another.

The solution or rather workaround to register a delegating OptingServlet, which holds references to a set of other servlets, and invokes them in turn isn't suitable for us, as it synchronizes too much development on multiple Sling-based applications by separate teams (distributed also geographically, but also organizationally).

Also workarounds for 2 servlets only doesn't work well in larger systems, as we have already encountered need for more than 2 OptingServlets. I guess you can imagine it being even harder when multiple development teams produce code containing OptingServlets which should just work depending on dynamically evaluated conditions and not globally dispatched or globally synchronized uniqueness of selectors.

Also the usage of regexp does not fulfil all our requirements. It also shows that the case of dynamic evaluation of conditions is not that very unusual as you might have assumed.

Additionally, the urge to limit OptingServlet's functionality for the sake of forcing _developers_ to use good practices of declarative configuration vs programatical opting for request handing really limits
platforms (even the larger ones) built on Sling, so that it is just not possible to use the platform for (yes, sometimes considered as edge-cases, but still needed in the real-live) solutions. I wouldn't like to support here the argument of developers facing a deadline to be valid in all cases. You cannot just to rule to prohibit selling kitchen knifes in order to prevent crimes...

OptingServlets - even if considered usable rather in exceptional cases, they have been introduced - that's the fact. So, they should also work (although it might not be a good practice to use it often) as intuitively as they are understood by merely reading [http://sling.apache.org/site/servlets.html#Servlets-OptingServletinterface] or [http://sling.apache.org/apidocs/sling5/org/apache/sling/api/servlets/OptingServlet.html].
                
      was (Author: krystian):
    The problem described in this issue doesn't seem to be an edge case for me. Please look at my example this time:

We have two (and possibly more) implementations of the OptingServlet interface. They are configured to work on the same resourceType, the same method and no selectors. Each of them chooses to accept a request based on a request parameter. For us they should be able to work independently as long as the accept logic for is mutually exclusive for both of them.
Sadly this is not the case. Only one of them is chosen to handle the request and if the selected servlet decides to not accept it, the other servlet is omitted. The resolver seems to choose the handling servlet based on an order determined while registering (starting bundle) - we are using Felix here.

We could of course choose to use different resourceTypes or selectors but this does not solve the problem, as there are multiple applications on the platform and each one of them can potentially break another.

The solution or rather workaround to register a delegating OptingServlet, which holds references to a set of other servlets, and invokes them in turn isn't suitable for us, as it synchronizes too much development on multiple Sling-based applications by separate teams (distributed also geographically, but also organizationally).

Also workarounds for 2 servlets only doesn't work well in larger systems, as we have already encountered need for more than 2 OptingServlets. I guess you can imagine it being even harder when multiple development teams produce code containing OptingServlets which should just work depending on dynamically evaluated conditions and not globally dispatched or globally synchronized uniqueness of selectors.

Also the usage of regexp does not fulfil all our requirements. It also shows that the case of dynamic evaluation of conditions is not that very unusual as you might have assumed.

Additionally, the urge to limit OptingServlet's functionality for the sake of forcing _developers_ to use good practices of declarative configuration vs programatical opting for request handing really limits
platforms (even the larger ones) built on Sling, so that it is just not possible to use the platform for (yes, sometimes considered as edge-cases, but still needed in the real-live) solutions. I wouldn't like to support here the argument of developers facing a deadline to be valid in all cases. You cannot just to rule not to prohibit selling kitchen knifes in order to prevent crimes...

OptingServlets - even if considered usable rather in exceptional cases, they have been introduced - that's the fact. So, they should also work (although it might not be a good practice to use it often) as intuitively as they are understood by merely reading [http://sling.apache.org/site/servlets.html#Servlets-OptingServletinterface] or [http://sling.apache.org/apidocs/sling5/org/apache/sling/api/servlets/OptingServlet.html].
                  
> Allow multiple OptingServletS for the same resource type
> --------------------------------------------------------
>
>                 Key: SLING-2191
>                 URL: https://issues.apache.org/jira/browse/SLING-2191
>             Project: Sling
>          Issue Type: Bug
>            Reporter: Reto Bachmann-Gmür
>         Attachments: SLING-2191-tests-launchpad.patch
>
>
> It should be possible to have multiple OptinServlets (accepting different sets of requests) for the same resource type.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira