18 messages in org.apache.wicket.usersRe: Same versioned link opens differe...| From | Sent On | Attachments |
|---|---|---|
| Alec Swan | Apr 11, 2012 2:36 pm | |
| Igor Vaynberg | Apr 11, 2012 2:55 pm | |
| Bertrand Guay-Paquet | Apr 11, 2012 3:07 pm | |
| Martin Grigorov | Apr 11, 2012 11:33 pm | |
| Alec Swan | Apr 12, 2012 8:22 am | |
| Igor Vaynberg | Apr 12, 2012 8:43 am | |
| Martin Grigorov | Apr 12, 2012 8:55 am | |
| Igor Vaynberg | Apr 12, 2012 8:57 am | |
| Alec Swan | Apr 12, 2012 9:32 am | |
| Bertrand Guay-Paquet | Apr 12, 2012 9:47 am | |
| Martin Grigorov | Apr 12, 2012 10:39 am | |
| Alec Swan | Apr 12, 2012 12:15 pm | |
| Dan Retzlaff | Apr 12, 2012 12:34 pm | |
| Igor Vaynberg | Apr 12, 2012 12:43 pm | |
| Bertrand Guay-Paquet | Apr 12, 2012 1:55 pm | |
| Igor Vaynberg | Apr 12, 2012 1:59 pm | |
| Martin Grigorov | Apr 13, 2012 12:54 am | |
| Alec Swan | Apr 13, 2012 12:38 pm |
| Subject: | Re: Same versioned link opens different pages on different machines | |
|---|---|---|
| From: | Alec Swan (alec...@gmail.com) | |
| Date: | Apr 12, 2012 12:15:23 pm | |
| List: | org.apache.wicket.users | |
"And I see no way to detect such problem by having just the information encoded in the url" This is VERY scary!
How do I fix this?
On Thu, Apr 12, 2012 at 11:39 AM, Martin Grigorov <mgri...@apache.org> wrote:
On Thu, Apr 12, 2012 at 7:47 PM, Bertrand Guay-Paquet <ber...@step.polymtl.ca> wrote:
I don't know much about HybridUrlCodingStrategy since I use Wicket 1.5, but based on what you observed (.x changes on every render), I would say the x is the page version.
I re-read your emails and if I understand correctly, both product 379 and 123 use the same page class. If that is the case, I don't think WICKET-4488 is the same problem. That issue concerns rendering a different *Page* class because of a page version id. In your case, the MyPage page class is the single page involved in both the wrong and right pages you see rendered.
^^This is what I tried to explain in my earlier mail.
user1 visits /product/10.8 then goes to /product/20.9 later user2 sends /product/30.8 to user1 user1 opens /product/30.8 but actually sees product/10 because Wicket loads page with id = 8 from the store ignoring the current request parameter (10)
So this is *not* covered by 4488! And I see no way to detect such problem by having just the information encoded in the url
I suggest putting a break point or logging code in your page's constructor to check 1) if a new page is created or an old version is displayed and 2) if the page parameters are properly decoded.
On 12/04/2012 12:32 PM, Alec Swan wrote:
"but because of the existence of a page with pageId 0 in the page store user sees page1, not page2 as user2 intended"
So, what is the page id in ../mp/oid/123.9 url?
On Thu, Apr 12, 2012 at 9:58 AM, Igor Vaynberg<igor...@gmail.com> wrote:
On Thu, Apr 12, 2012 at 8:55 AM, Martin Grigorov<mgri...@apache.org> wrote:
On Thu, Apr 12, 2012 at 6:43 PM, Igor Vaynberg<igor...@gmail.com> wrote:
On Thu, Apr 12, 2012 at 8:22 AM, Alec Swan<alec...@gmail.com> wrote:
Igor,
The link I click ends with /mp/oid/123.9, where 123 is a product id. However, when the page is rendered its URL changes to end with /mp/oid/123.x where x is different every time. Moreover, the page is displaying the wrong product 379!
So, it's not the wrong version of the page, but the wrong product that worries me.
Can you explain this?
i never once said the word "version" in my response :) i was, in fact, talking about page ids. i dont think the old hybrid url coding strategy checks the page id in its url against the mount. so in your session page 9 can be something entirely different. i believe this is what was fixed by WICKET-4488.
Not exactly. 4488 fixes the problem when user1 has opened page1?0. Then user2 sends a url page2?0 to user1. user1 opens this url but because of the existence of a page with pageId 0 in the page store user sees page1, not page2 as user2 intended.
After 4488 Wicket will load page with id == 0 then it will check its class against the class of the mountPoint and ignore the stored page if they don't match. And will create a completely new instance of mountPoint and show it to user1. I.e. a new instance of page2
I hope I described it clearly.
i think thats *exactly* what i said...
-igor
Thanks,
Alec
On Wed, Apr 11, 2012 at 4:08 PM, Bertrand Guay-Paquet <ber...@step.polymtl.ca> wrote:
Hi,
A ticket regarding this was created and resolved in 1.5 (WICKET-4488). From the work log: "There was code for this situation but it didn't cover the case 100%. Now if a request to page2?0 is made and the type of the found page with id=0 is not Page2 then a new instance of Page2 is instantiated."
On 11/04/2012 5:56 PM, Igor Vaynberg wrote:
page 5 in your session can be completely different then page 5 in user's session.
non-bookmarkable urls cannot be emailed...thats kind of the point.
-igor
On Wed, Apr 11, 2012 at 2:37 PM, Alec Swan<alec...@gmail.com> wrote:
Hello,
I received a link from a customer to a versioned page (.version at the end of the URL). However, when I click on the link I see a completely different page.
We are using Wicket 1.4.17 and the page is mounted as:
mount(new HybridUrlCodingStrategy("mp", MyPage.class));
Why is this happening and how can I fix this?
Thanks,
Alec
-- Martin Grigorov jWeekend Training, Consulting, Development http://jWeekend.com
-- Martin Grigorov jWeekend Training, Consulting, Development http://jWeekend.com