[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [virgo-dev] [snaps] what's next?

Hi Dmitry

I don't have any strong opinions, but I've drawn together the list of relevant SpringSource JIRAs [1] for anyone who wants to read them and I've added a pointer to this thread to those issues which were raised by users (rather than SpringSource engineers).


[1] https://issuetracker.springsource.com/secure/IssueNavigator.jspa?mode=hide&requestId=11792

On 22 Jul 2010, at 15:18, Dmitry Sklyut wrote:

Hi All,

I wonder what else can be done with spans?  What was left in the SpringSource jira with regards to slices?  Should any of that stuff be examined for inclusion?

>From my view point, there are few things that I would like to have:

1. Well known strategy for including/excluding servlet context/session/request attributes.
For example, we ran into an issue with both host and snap as Spring MVC apps.  Both would try to add their SpringMVC attributes to the servlet context/request attributes.  Things like bellow:
public static final String HANDLER_EXECUTION_CHAIN_ATTRIBUTE = DispatcherServlet.class.getName() + ".HANDLER";
public static final String WEB_APPLICATION_CONTEXT_ATTRIBUTE = DispatcherServlet.class.getName() + ".CONTEXT";
public static final String LOCALE_RESOLVER_ATTRIBUTE = DispatcherServlet.class.getName() + ".LOCALE_RESOLVER";
public static final String THEME_RESOLVER_ATTRIBUTE = DispatcherServlet.class.getName() + ".THEME_RESOLVER";
public static final String THEME_SOURCE_ATTRIBUTE = DispatcherServlet.class.getName() + ".THEME_SOURCE";

I fairly sure other frameworks make assumptions based on some conventions.

2. Allowing snaps placing stuff in a session to be visible to host.  That is sharing session space between host and snap(s)

3. Lock down semantics of request path handling/parsing.  This is kind of related to https://issuetracker.springsource.com/browse/DMS-2311.

All of these ideas will need some research.  For example - do snaps allow placing of stuff in servlet context/session if host does not have visibility into types?  Should it be an error or a warning?
URL/path handling will need to define very clear semantics.  Should snaps always treat snap as an independent web-app and decorate request in a way that consistent with a snap as a web module?

Are there any other wishes and ideas that users have?