Bug 165630 – Add 64-bit signed LEB decode method

• Component: New Bugs
• Status: RESOLVED FIXED

[Keith Miller]

Add 64-bit signed LEB decode method

[Keith Miller]

Created attachment 296576 [details]
Patch

[Keith Miller]

Created attachment 296591 [details]
Patch

[Ryosuke Niwa]

Comment on attachment 296576 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=296576&action=review

> Source/WTF/wtf/LEBDecoder.h:59
> +template<size_t maxByteLength, typename T>
> +inline bool WARN_UNUSED_RETURN decodeInt(const uint8_t* bytes, size_t length, size_t& offset, T& result)

I think it would have been better maxByteLength was obtained via traits on T.
e.g.
template <typename T> struct maxLEBBytes;
template <> maxLEBBytes<uint32_t> struct { static size_t maxByteLength = 5; }
template <> maxLEBBytes<unit64_t> struct { static size_t maxByteLength = 10; }
This would be less error prone than passing in matching template arguments.

> Source/WTF/wtf/LEBDecoder.h:65
> -    size_t last = std::min(max32BitLEBByteLength, length - offset - 1);
> +    size_t last = std::min(maxByteLength, length - offset) - 1;

You should mention -1 moving out of std::min in the change log.

[Ryosuke Niwa]

Comment on attachment 296591 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=296591&action=review

> Source/WTF/wtf/LEBDecoder.h:41
> +    const size_t maxByteLength = (numBits - 1) / 7 + 1; // numBits / 7 rounding up.

Nice!

[Keith Miller]

Committed r209586: <http://trac.webkit.org/changeset/209586>