Bug 87811 – correct the sequence of diagnostic crash check

• Status: RESOLVED
• Resolution: WONTFIX
• Component: New Bugs

[Lu Guanqun]

correct the sequence of diagnostic crash check

[Lu Guanqun]

Created attachment 144678 [details]
Patch

[Lu Guanqun]

We should do the null check before we de-reference the pointer. Otherwise, we don't get the benefits of crash check.

[James Robinson]

Comment on attachment 144678 [details]
Patch

I think you slightly misunderstand the meaning of the tileAt() check - we're trying to verify that _after_ adding the tile it exists.  Moving the CHECK() for addedTile seems reasonable.

[Lu Guanqun]

(In reply to comment #3)
> (From update of attachment 144678 [details])
> I think you slightly misunderstand the meaning of the tileAt() check - we're trying to verify that _after_ adding the tile it exists.  Moving the CHECK() for addedTile seems reasonable.

It does check the tile _after_ it's been added, since it's after the line "m_tiler->addTile(tile.release(), i, j);".

The line "addedTile->dirtyRect = m_tiler->tileRect(addedTile);" doesn't add the tile, does it?

Do I miss something here?

[James Robinson]

Comment on attachment 144678 [details]
Patch

Ah sorry, you're right.  Still not sure it's worth churning the line numbers, we're going to remove these lines in a few days anyway.

[Lu Guanqun]

(In reply to comment #5)
> (From update of attachment 144678 [details])
> Ah sorry, you're right.  Still not sure it's worth churning the line numbers, we're going to remove these lines in a few days anyway.

OK. It's good to know that these lines would be removed. So I will mark this bug as won't fix.

Thanks for the review anyway.