382702 - Nonexistant about: url should show an error page rather than throwing a dialog

Status: RESOLVED FIXED

(Core :: DOM: Navigation, defect)

Description

[Gábor Stefanik]

Currently, if the user attempts to go to a nonexistent about: URL (like about:nothing), he/she is greeted with an intrusive error dialog. I suspect that the same happens if a FRAME, OBJECT or IFRAME attempts to load such a page. This is definitely unacceptable. We need error pages, like for the rest of the errors.

Comment 1

[Gábor Stefanik]

->All/All.

Comment 2

[Adam Guthrie]

This is actually a docshell bug: http://bonsai.mozilla.org/cvsblame.cgi?file=/mozilla/docshell/base/nsDocShell.cpp&rev=1.834&mark=2988-2990#2986

ref. bug 291876 and bug 306409.

Comment 4

[Boris Zbarsky [:bzbarsky]]

We already have a bug on the "use error page when we can't create URI" problem, no?  This would be a duplicate.

Comment 5

[O. Atsushi (Torisugari)]

(In reply to Boris Zbarsky (:bz) from comment #4)
> We already have a bug on the "use error page when we can't create URI"
> problem, no?  This would be a duplicate.

I can't find another "backout of bug 291876" bug.

How about setting "about:blank"? i.e. for an "about:nothing" request,

docshell::currentURI:
  "about:blank"

onLocationChange:
  "about:blank"

Session history:
  "about:blank"

Global history / bookmark:
  kick.

content.location.href:
  "about:blank"

contentDocument.documentURI:
  "about:neterror?e=malformedURI&u=about:nothing&d=Hey,%20watch%20out!"

URLbar:
 "about:blank" or "about:nothing"

The session history will work with this kind of hack?

Comment 6

[O. Atsushi (Torisugari)]

Attachment: proposal v1

It seems to me the session history works as expected.

Comment 7

[Christian :Biesinger (don't email me, ping me on IRC)]

Comment on attachment 581971 [details] [diff] [review]
proposal v1

Interesting. Yeah, this should work, and it seems reasonable to me. One thing, do add a comment above your NS_NewURI call describing why you're adding this.

Comment 8

[O. Atsushi (Torisugari)]

Oh.

I don't have enough time to write a new patch for a month or two.

(In reply to Christian :Biesinger (don't email me, ping me on IRC) from comment #7)
> do add a comment above your NS_NewURI call describing why you're
> adding this.

Then, something like this:
+// Something is wrong with converting from |aURL| to |aURI|. We don't like to let |mFailedURI| NULL (see bug 291876), we provisionally set "about:blank" instead.

Comment 9

[Josh Matthews [:jdm]]

If Christian signs off on a particular comment, I'm happy to update the patch and land it.

Comment 10

[O. Atsushi (Torisugari)]

(In reply to Josh Matthews [:jdm] from comment #9)
> If Christian signs off on a particular comment, I'm happy to update the
> patch and land it.

Thank you very much!

Maybe you've already realized or not, this patch heavily (?) depends on bug 312680's landing. And, btw, I should have put |rv| outside of the {} block and have made it reusable if that is not too confusing to other developers.

Comment 11

[Christian :Biesinger (don't email me, ping me on IRC)]

Comment on attachment 581971 [details] [diff] [review]
proposal v1

I'd go for:

// We need a URI object to store a session history entry, so make up a URI

Comment 12

[Josh Matthews [:jdm]]

http://hg.mozilla.org/integration/mozilla-inbound/rev/1dcee8f1559a

Comment 13

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/1dcee8f1559a

Comment 14

[henryfhchan]

in one of the older bugs, restoring about:blank on session restore caused a large performance hit.  was this fixed, or does the fix for this bug cause this problem again?