#194 refresh metatag redirects immediately

[yhal003]

code:
public static void main(String[] args) {
try{
final String url = "http://localhost:8080/Hello.html";
final WebClient client = new WebClient();
HtmlPage page = (HtmlPage)client.getPage(new URL(url));
System.out.println(page.asText());
}
catch(Exception ex){}
}

with html pages:
Hello.html

<meta http-equiv="Refresh" content="500000;url=/redirect.html"> hello

redirect.html

redirect

when it runs I get page = redirect.html and Hello.html
is "lost"
althought redirect should follow after 50000 sec;
is that bug?

[Brad Clarke]

Logged In: YES
user_id=257129

meta redirects are processed immediately but this should
probably be an optiona thing

[Daniel Gredler]

Logged In: YES
user_id=1109422

According to every page I've seen (for example,
http://www.htmlhelp.com/reference/html40/head/meta.html),
you're supposed to be able to specify the time to wait in
seconds. I think HtmlUnit's behavior is flat out incorrect
in executing the refresh immediately.

I've been looking in the code, and don't really understand
the need for RefreshHandler and DefaultRefreshHandler. Why
would someone want to implement a custom RefreshHandler? Why
does DefaultRefreshHandler ignore the timeBeforeRefresh
parameter? The JavaDocs for these handlers mention that
refreshes can be triggered via JavaScript, but the handlers
are only used for refreshes triggered via response headers
or meta tags. Can the refresh handlers be taken out?

[yhal003]

Logged In: YES
user_id=1191274

Well that is what I did to resolve that issue - implemented
Refresh handler so that it never redirects-

package com.gargoylesoftware.htmlunit;
import java.net.URL;

public class FakeHandler implements RefreshHandler{
public boolean shouldRefresh(Page page,URL url,int integer){
return false;
}
}

and then assigned it to the client:

client.setRefreshHandler(new FakeHandler());

[Marc Guillemot]

Logged In: YES
user_id=402164

I'm not the author of the refersh handler related code but I
understand the implemenation: it is simple and safe.
Waiting for the indicated delay would require to start a
separate thread that would perform the refresh after the
indicated time. In the meantime, it should be possible to
manipulate the page containing the <meta refresh="">, and
during this manipulation, the page could "lose its window".
If the javadoc is wrong, then it should be fixed.
I think that it is usefull to have the possibility to
specify the way refresh should be performed: in most cases
an immediate refresh allows tests to run faster, but
sometime the business logic requires to wait for the
specified time. Optimal would be to give the possibility to
work like browsers do as well, taking care of the problems
I've mentionned before.

Perhaps could it be convenient to provide some
RefreshHandler instances:

• a "no refresh" RefeshHandler excactly like yha1003's
  FakeHandler dismissing refresh requests.
• a "wait refresh" RefreshHandler performing a wait for the
  time.

I think that this bug can be closed. Please reopen if it is
not the case.