Re: [webkit-dev] Proposal: Not supporting x86 w/o SSE2

2018-08-04 Thread Michael Catanzaro
Technically, many of WebKitGTK+'s distributors require that it not use SSE2 instructions in 32-bit builds. I am no longer sure which ones. (It used to be the case for Fedora, for example, but that changed recently.) In practice, so few people care about these old machines anymore that such

Re: [webkit-dev] Proposal: Not supporting x86 w/o SSE2

2018-07-29 Thread Yusuke SUZUKI
On Mon, Jul 30, 2018 at 4:37 AM Darin Adler wrote: > Sounds good. > > There are clients of WebKit outside web browsing. A significant client > like that on Windows at Apple is iTunes. I checked < > https://www.apple.com/itunes/download/> and Windows versions of iTunes > require a processor with

Re: [webkit-dev] Proposal: Not supporting x86 w/o SSE2

2018-07-29 Thread Darin Adler
Sounds good. There are clients of WebKit outside web browsing. A significant client like that on Windows at Apple is iTunes. I checked and Windows versions of iTunes require a processor with support for SSE2, so clarifying WebKit’s lack of support won’t

[webkit-dev] Proposal: Not supporting x86 w/o SSE2

2018-07-29 Thread Yusuke SUZUKI
Hello WebKittens, To mitigate recent side-channel attacks, we added `lfence` in x86 WebKit. But this is supported in x86 SSE2. So the ToT WebKit is not usable with very old x86 CPUs which do not have SSE2[1]. According to [2], Mozilla Firefox 49 no longer supports x86 w/o SSE2. Since Firefox 45