CODEWITHSUNDEEP
wicket
Jan Krakora
Jan Krakora

Reputation: 2610

Wicket page versioning and history support

Could anyone explain me what is the Wicket's page versioning useful for? There is an article in the FAQ that is related to this topic:

Wicket stores versions of pages to support the browser's back button.

Suppose you have a paging ListView with links in the ListItems, and you've clicked through to display the third page of items. On the third page, you click the link to view the details page for that item. Now, the currently available state on the server is that you were on page 3 when you clicked the link. Then you click the browser's back button twice (i.e. back to list page 3, then back to list page 2, but all in the browser). While you're on page 2, the server state is that you're on page 3. Without versioning, clicking on a ListItem link on page 2 would actually take you to the details page for an item on page 3.

But unfortunately I don't understand it at all. When I click on a ListItem on page 2, I would expect to get to the page defined by that Link - details page for the item. Why should I get on the details page of the item on page 3?

Moreover, when one press the back button in a browser, it doesn't call the server at all. Is it right?

So how this versioning works?

Upvotes: 1

Views: 1483

Answers (1)

tetsuo
tetsuo

Reputation: 10896

No, the server is not notified when you press the back button. I'll try to explain what happens in the example:

  • You access you application for the first time. On the server, a page 'list' is created, used to render the HTML you see in your browser, and stored as page v1. You see the first 10 items of the list.

  • You click the 'next' link, and it refers to a link in page v1. On the server, page v1 is loaded, the link logic is executed (to advance the pagination), the page is used to render HTML, and is stored as page v2. You see items from 11 to 20.

  • You click the 'next' link, and it refers to a link in page v2. On the server, page v2 is loaded, the link logic is executed (to advance the pagination), the page is used to render HTML, and is stored as page v3. You see items from 21 to 30.

  • You click the 'details' link for item 25, and it refers to the link for the 5th item in page v3 (this page only shows 10 items, and the link, even if it refers to the 25th item in the complete list, in this page it's just the 5th). On the server, page v3 is loaded, its logic is executed, page 'detail' is created, stored as page v4, and you are redirected to it. Your browser requests page v4, the server loads it, and uses it to render your HTML page (no new version is stored, since it's just rendering). You see the details for item 25.

  • You click the browser's 'back' button 2 times, and see page 'list' showing items 11 to 20, referring to page v2 (list). Then you click a link 'details' for item 13. On the server, page v2 is loaded (not v4, the last one executed), since the link clicked pointed to this page version. Then, the 3rd item link's logic is executed, a new page 'details' is created, stored as page v5, and you are redirected to it. The browser requests page v5, the server loads it, and uses it to render your HTML. You see the details for item 3.

All this may seem strange if you come from a Struts-like background, where you always just put the item id or which page to show as a link parameter. In Wicket, the usual case is to store all state in the server, and navigation is not done by the client (direct link to another page passing parameters), but in the server. A link just asks the server to execute code in a page object version, the navigation is done all server-side.

You could argue that the Struts style is simpler (and you can do it in Wicket too, it just isn't optimal), but keeping the state only in the server has many advantages. Fist, once you get used to it, it's actually much easier. No need to add every single param to a pagination link (search parameters, first item, page length, sorting column, order direction, etc.). Also, you avoid many security issues (you can't just change the URL id param to an arbitrary value and access other users' data), and can control everything from Java code instead of mixed Java-Javascript (you still can do it if you want, though).

Upvotes: 4

Related Questions