Our great sponsors
-
WebKit
Home of the WebKit project, the browser engine used by Safari, Mail, App Store and many other applications on macOS, iOS and Linux.
-
-
SurveyJS
Open-Source JSON Form Builder to Create Dynamic Forms Right in Your App. With SurveyJS form UI libraries, you can build and style forms in a fully-integrated drag & drop form builder, render them in your JS app, and store form submission data in any backend, inc. PHP, ASP.NET Core, and Node.js.
-
-
To use this in Bun, you’d have to start Bun with the environment variable “BUN_JSC_useDollarVM=1” and then $vm.createBuiltin(mySourceCodeString)
When using this intrinsic, if any of the arguments are incorrect or it cannot otherwise enable it, the entire process will probably crash. In debug builds of JSC it will have a nicer assertion failure but that is not enabled in release builds
Example code: https://github.com/WebKit/WebKit/blob/17351231b4dedb62d81721...
also happy to answer any questions about Bun
This is not actually about Tail Call Optimisation, which is more flexible and optional matter of optimisation, but about Proper Tail Calls, which are actually part of the ECMAScript 6 specification (over implementer objections)—in strict mode, calls in tail position must not create additional stack frames. This is the last piece of ECMAScript 6 that most engines haven’t implemented, because it’s rather controversial: it actually causes some performance problems, and makes debugging harder, and may have security issues (in 2016, Mozilla declared it impossible to implement across realm boundaries due to their security model).
https://github.com/tc39/proposal-ptc-syntax has a lot of useful information about it all, and a proposal to make it explicit in syntax, such as with `return continue …`.
(Fun terminology problems here. The term TCO is commonly used for PTC, and PTC is very close to being a subset of TCO, but the mandatory stack frame elision which ruins debugging feels to me like it falls outside of TCO. In various situations, debuggers will mark things like “stack frame omitted” when they’ve optimised one out of existence, but you can generally compile things differently, or something like that, to prevent this. But with PTC, it feels like the engine is kinda not even allowed to know that a stack frames may be absent. So I say PTC and TCO are a little distinct, though PTC is mostly just a subset of TCO. Reminds me of the terminology of tree-shaking versus dead code removal—where the former is essentially a subset of the latter, but that the effects are just slightly different, though I’d say it’s more slight in that case than this.)
While working on effection (https://github.com/thefrontside/effection) we spent a bunch of time ensuring that our delimited continuations could handle deep recursive call stacks in Deno.
PR: https://github.com/thefrontside/continuation/pull/11
TCO would have definitely simplified this issue.
What’s worse is hitting maximum memory callstack exception is very tricky to catch and is not reliable across runtimes. So when a user hits it it can be tricky to track down.
While working on effection (https://github.com/thefrontside/effection) we spent a bunch of time ensuring that our delimited continuations could handle deep recursive call stacks in Deno.
PR: https://github.com/thefrontside/continuation/pull/11
TCO would have definitely simplified this issue.
What’s worse is hitting maximum memory callstack exception is very tricky to catch and is not reliable across runtimes. So when a user hits it it can be tricky to track down.