1543943 - Port Bug 1543564 part 3. Remove use of nsPIDOMWindowOuter::GetOuterWindow()

Status: RESOLVED FIXED

(MailNews Core :: General, task)

Description

[Jorg K (CEST = GMT+2)]

https://searchfox.org/comm-central/search?q=GetOuterWindow&case=false&regexp=false&path=mailnews

Four call sites.

Comment 1

[Jorg K (CEST = GMT+2)]

Attachment: 1543943-GetOuterWindow.patch

This should do it.

Comment 2

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/comm-central/rev/cd367a28aeb9
Port bug 1543564, part 3: Remove use of nsPIDOMWindowOuter::GetOuterWindow(). rs=bustage-fix DONTBUILD

Comment 3

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 9057825 [details] [diff] [review]
1543943-GetOuterWindow.patch

r=me, but this is leaving some unneeded NS_ENSURE_ARG_POINTER bits around, right?

Comment 4

[Jorg K (CEST = GMT+2)]

I asked myself that question. The code pattern is this:

1    nsCOMPtr<nsPIDOMWindowOuter> parentWindow = nsPIDOMWindowOuter::From(domWindow);
2 -  parentWindow = parentWindow->GetOuterWindow();
3    NS_ENSURE_ARG_POINTER(parentWindow);

So surely, after line 1 `parentWindow` was never null, otherwise line 2 would have crashed. If the behaviour of line 1 hasn't changed, then line 3 can go after line 2 is gone. I left them in for now, since they don't hurt.

I'm not really familiar with what's going on here, so if your suggestion is to remove the checks, I'll remove them. Please let me know.

Comment 5

[Boris Zbarsky [:bzbarsky]]

So surely, after line 1 parentWindow was never null

Correct.

then line 3 can go after line 2 is gone

Yep. Removing them makes the code clearer and a tiny bit faster...

Comment 6

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/comm-central/rev/e1c03b7fdd86
Follow-up: Remove needless null checks as suggested by :bz. r=me