|Written by Kay Ewbank|
|Monday, 12 August 2019|
The developers have overhauled the JSON parser to be much faster at scanning and parsing JSON, resulting in parsing of data served by popular web pages being handled up to 2.7 times faster. Until this release, the JSON parser was recursive and used native stack space relative to the nesting depth of the incoming JSON data, so could run out of stack for very deeply nested JSON data. This has been replaced by an iterative parser that manages its own stack and is limited only by available memory. The parser has also been reworked to be more memory-efficient.
The improvements follow other performance upgrades to JSON parsing added in the previous two point releases, 7.4 and 7.5. 7.4 improved the handling of calls with too few or too many parameters, in some cases reducing the call overhead by 60 percent.
Another improvement from 7.4 was a fix to handle large scripts in Chrome better, dealing with UTF-8 decoding and property names more appropriately to provide a combined 18 percent improvement.
Promise handling has been improved with the addition of promise.allsettled. This provides a signal when all the input promises are either fulfilled or rejected, so settled. The developers say this is useful in cases where you don’t care about the state of the promise, you just want to know when the work is done, regardless of whether it was successful.
Native stack walking support has been added for Windows. Until now, Windows couldn't walk a call stack containing code generated by TurboFan when running on the x64 architecture. V8 can now provide Windows with the metadata it needs be able to walk these stacks on x64, and in v7.6 this is enabled by default.
or email your comment to: email@example.com
|Last Updated ( Monday, 12 August 2019 )|