The ability to disable the REFRESH meta tag, which can redirect the document to another page, is needed.

The SH related bug on this 39938 has been fixed. There is a mechanism in place to identify META REFRESH based redirections. Look in to nsDocShell::refreshURI(), nsRefreshTimer::Notify() in nsDocshell.cpp, and layout/html/document/src/nsHTMLContentSink.cpp

Radha, I'm not following. The changes in 39938 apply to manipluating whether or not redirects go into session history. We need a way to expose disabling of these redirects.

I will try to implement this for mozilla 0.9. I'm presuming that we can go thro' some Prefs to achieve this.

We'll need this on a top-level window basis, and prefs are global. For example a mail viewer won't want to support this, but browser windows would. Scott, how does mail/news do this?

this should be a setting on nsIWebBrowserSetup. maybe something similair to http://lxr.mozilla.org/mozilla/source/editor/base/nsEditorShell.cpp#5281 can be done?

"refresh" is too generic a term. It can translate to HTTP status code refreshing, javascript refreshing, and META refreshing. I propose renaming if we just want to block META refreshing, then we should rename the attribute in nsIDocShell to "allowMETARedirects" and the webbrowsersetup code should be renamed to SETUP_ALLOW_META_REDIRECTS . With that said however, would it be much work indeed make it generic enough to block both META *and* javascript redirects? If not, then we should keep your naming convention, and modify the implementation to block both. What do you think?

hey radha, I took a look at the patch and it looks good... I think that you should move the test of GetAllowRefreshRedirects(...) up to the top of the method (say right after checking mDocShell)... Because, it looks like now, the flag is only being checked if the refresh URI is different from the current one - so meta-refreshes to the same URI will *not* be disabled... Does this sound right? -- rick

reply to the question: We definitely want to disable META refreshes that re-direct, but we can keep them if they are the same URL (ie. Anchors, etc).

Based on marek's comments, the original patch that disables *only* the redirects stays. I have however renamed the method name in docshell and the flag name in nsiWebBrowserSetup.idl as per jud's comments. The JS issue is addressed in a different bug. Can I get a r and sr? Thanks

r=valeski

If we are going to allow meta-refreshes to the same URI, then we may need another flag which disables *all* refreshes... I assume that we will want this for the editor case, where we don't want a document with a meta-refresh to fire while it is being edited... Can we add an argument to AllowRefreshRedirects(...) indicating whether all redirects, or just "foriegn" ones should be disabled? I guess I'm also a bit confused as to why redirects to the same URI are "ok"... -- rick

This was fixed last week.

Correction: Changing QA contact for the Embed API bugs to David Epstein.

SETUP_ALLOW_META_REDIRECTS flag declared in nsIWebBrowserSetup, used in nsWebBrowser.cpp (calls mDocShell->SetAllowMetaRedirects(aValue)).

Sorry to ask a stupid question, but what is the name of the preference that I need to add to user.js (or about:config) in order to activate this fix and disable META refresh? I've tried everything I can think of, and no go. Thanks.

It looks like the fix vanished without a trace soon after it was put in. It has cropped up again as bug 83265.

This bug added an embedding API to disable meta-refresh. The code is still there and presumably still works, but our own browser does not ever set that property.

> "refresh" is too generic a term. It can translate to HTTP status code > refreshing, javascript refreshing, and META refreshing. I take it, this API function only covers "META refreshs". Are there already API calls available for blocking "Javascript refreshs" and "HTTP 301 Redirects"?