1393196 - gfx/ has lots of signed/unsigned comparisons that generate warnings

Status: RESOLVED WONTFIX

(Core :: Graphics, defect, P3)

Description

[Tom Ritter [:tjr] (OOTO until 4/30 at least)]

In many places in gfx/ we compare unsigned and signed ints.

I'm compiling for Windows, on Linux (Bug 1330608) so it's exposing lots of warnings. I'm trying to understand how many of these we care about (if any) and if they are brand-new or if we had previously seen them and chosen to disable the warning or something.

One class I identified is we have a DXGI_SWAP_CHAIN_DESC object (https://msdn.microsoft.com/en-us/library/windows/desktop/bb173075(v=vs.85).aspx ) with a submember of type DXGI_MODE_DESC (https://msdn.microsoft.com/en-us/library/windows/desktop/bb173064(v=vs.85).aspx ).  We often compare this object to a Mozilla type: LayoutDeviceIntSize which is signed. I've attached a patch that shows several of these sites (with a naive attempt at safely resolving them.)

Comment 1

[Tom Ritter [:tjr] (OOTO until 4/30 at least)]

Attachment: Bug 1393196 Some illustrative examples of gfx/ signed/unsigned comparisons

Review commit: https://reviewboard.mozilla.org/r/171786/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/171786/

Comment 2

[Jeff Muizelaar [:jrmuizel]]

We should probably be more deliberate about what we do here.

Comment 3

[Tom Ritter [:tjr] (OOTO until 4/30 at least)]

Compiling with gcc for Windows is not supported anymore, so this is a WONTFIX