All of these places where you throw new Error can be removed if you return the error from the op.

I think it’s better not to panic - because users can still trigger that by calling the op directly. Better to return Err(...)

I’m not at my computer to give a good example but you should be able to look through the other ops and find an example of returning an error.

const response = await fetch(new URL('http://www.example.com/démonstration.html'));

BTW, the URL implementation does not meet the specification

ref: https://url.spec.whatwg.org/#api https://w3c.github.io/FileAPI/#creating-revoking

<!-- Before submitting a PR read https://deno.land/manual.html#contributing -->

It seems you could reduce code here by lifting the unwrap_or_default().into_os_string()... into the next statement.

This seems to work fine, but it's not exactly ideal. The rust compiler doesn't allow me to be more implicit here sadly.

ref #51

What if we land this PR but add a TODO to replace the implementation with URL.searchParams later? What do you think @justjavac ?

<!-- Before submitting a PR read https://deno.land/manual.html#contributing -->

upgrade to v0.26.0 until there is a viable synchronization solution

@lucacasonato that is the eventual goal. In fact the original patches from @afinch7 were doing something along those lines IIRC. However we need some foundational infrastructure - having basic dlopen functionality is important and close to the OS. We will add sugar on top of that.

Fixes #60.

(I had actually already run into this and changed it locally!)

When trying to build with missing submodules(I didn't add --recursive to the submodule init command) it will still give build success. This is on ubuntu 18.04 not sure if this can be replicated on other platforms.

Closes #76.

I think what @justjavac is suggesting is that we first implement the query string part of the URL spec in Deno. Then std/node/qs can use that implementation, rather than having a completely separate one.

Unfortunately you're probably right... I think the Stream API is unnecessarily complicated. I've presented my alternative version of streaming by porting Golang's I/O system, which include things like Deno.copy. That said, users don't care, they just want browser-like...

This is certainly an improvement - but please add a TODO to make it possible for targets other than Local<Value>.. For example Local<String> should be able to be down cast to Local<Name>.

The Unexpected entry was not being rendered because the type of the entry is "esm" not "github" and we provide no valid repo. Fix the repo/owner properties in the database but, since there are no useful files in the repo wire up displaying README.md in our index page.

While here I also discovered that any missing type would actually cause an exception because it fell off the end of renderDir(). This case was actually covered in the older registry repo, so port that.

ref: #3483

Make sure exists() doesn't swallow permissions errors

• [x] test case

Add library for running SQLite from Deno: https://github.com/dyedgreen/deno-sqlite

Now we get the file and fileInfo sequentially. This change is to make it concurrently, which reduces time cost from open() + stat() to max(open(), stat())

Please maintain original copyright headers

Looks good to me...

<!-- Before submitting a PR read https://deno.land/manual.html#contributing --> cc @kitsonk close: #3465

/cc @zhmushan

$ deno --version
deno 0.26.0
$ deno --allow-net https://deno.land/std/http/file_server.ts
HTTP server listening on http://0.0.0.0:4500/

access server via browser or curl and got 500 error

$ curl http://0.0.0.0:4500/
Internal server error

Align deno_http and node_http test server response to have similar headers and sizes (s.t. benchmark reflects more correctly).

After this change: curl -i on Node:

HTTP/1.1 200 OK
Date: Thu, 12 Dec 2019 01:31:15 GMT
Connection: keep-alive
Content-Length: 11

Hello World

curl -i on Deno:

HTTP/1.1 200 OK
date: Thu, 12 Dec 2019 01:30:58 GMT
connection: keep-alive
content-length: 11

Hello World

fixes the manual for: https://github.com/denoland/deno/pull/3389

removed the gist link https://gist.github.com/ry/a610ce48cba2f0225f9c81a5a833fc87 as it points to something else.

took me a while to figure out on how to apply an option or multiple options, since there is no feedback when applied wrong.

$ deno --v8-flags=foo-bar  # no error
$ deno --v8-flags=--foo-bar  # no error

Hi @zhmushan - there's a conflict, could you please merge master? Otherwise looks good. Thank you!

Previously we did not handle automatic connection closing on .respond() error. This leaves the supposedly dead connection not closed (CLOSE_WAIT) (since req.done is never resolved due to error), resulting in errors like Too many files open. (which might be the cause of https://github.com/denoland/deno/issues/3463#issuecomment-563056609)

A simple demonstration of what this patch fixes is with the following:

import { serve } from 'https://deno.land/std/http/server.ts';
const body = new TextEncoder().encode("Hello World\n");
const s = serve({ port: 8000 });

console.log("Listening on port 8000");

setInterval(() => {
  console.log(Object.keys(Deno.resources()).length);
}, 10000);

for await (const req of s) {
  req.respond({ body }).catch(() => {});
}

If you fire a few autocannon -c 100 -d 5 -p 10 http://localhost:8000 to this, you'll notice that Object.keys(Deno.resources()).length keeps growing. (likely due to autocannon closing conn on its side thus making .respond() fail (especially under pipelining))

With this patch, Object.keys(Deno.resources()).length should eventually return to a fixed amount.

Looks good - thank you! Sorry we had a problem with windows build being flaky - could you please merge master branch?

Currently if you run deno https://deno.land/std@v0.26.0/testing/runner.ts without sufficient permissions it'll just report No matching test modules found..

It seems this little 3 line bit of code is repeated many times. Should it be abstracted into a function?

Kevin - yes that sounds good .. please leave a Todo comment explaining the situation

For some reason when using stdout the permissions are stripped.

Fixes #3461.

<!-- Before submitting a PR read https://deno.land/manual.html#contributing -->

$ ls -lh
-rw-r--r--@  1 axetroy  staff    43M 12  8 02:18 deno_osx_x64
-rw-r--r--@  1 axetroy  staff    11M 12  8 02:17 deno_osx_x64.gz
$ ./deno_osx_x64 --version
fish: The file './deno_osx_x64' is not executable by this user
$ chmod +x deno_osx_x64
$ ./deno_osx_x64 --version
deno 0.26.0
$ ls -lh
-rwxr-xr-x@  1 axetroy  staff    43M 12  8 02:18 deno_osx_x64
-rw-r--r--@  1 axetroy  staff    11M 12  8 02:17 deno_osx_x64.gz

@lucacasonato Oh - yes that's what I thought. I don't use Windows so based on the above comments I thought maybe tar support was added. I'm fine with switching to zip for everyone. It's annoying to have two types of packages.

why remove this?

This PR adds rand module for easy access to random numbers:

• https://github.com/rsp/deno-rand

Hey its my first pr to open source project, let me know if I have to do some additional stuff. Was not able to test it with actual files, but now if you go to 0.0.0.0:4500/test test.txt fsPath will be with space not %20.

Fixes #3164 <!-- Before submitting a PR read https://deno.land/manual.html#contributing -->

https://deno.land/std/http/file_server.ts can not handle urls like /test%20file.txt - it does not seem to parse URL encoded characters like %20.

Hmm even FOLDERID_Profile isn't set Deno shouldn't panic.

Can you pull the constructors out of the unsafe block?

Also rather than // TODO: safer casts put

// TODO Safer casts https://github.com/denoland/rusty_v8/issues/51

<!-- Before submitting a PR read https://deno.land/manual.html#contributing -->

I don't think the tests should depend on curl. Can you use fetch instead?

We could require that script args are placed after -- so it would be

deno -A https://deno.land/std/http/file_server.ts -- /etc

This would allow for better error checking of the flags...

https://github.com/denoland/deno/pull/3470/checks?check_run_id=340389213

2019-12-09T16:52:40.7820723Z test basic ... FAILED
2019-12-09T16:52:40.7820950Z 
2019-12-09T16:52:40.7821938Z failures:
2019-12-09T16:52:40.7822131Z 
2019-12-09T16:52:40.7822526Z ---- basic stdout ----
2019-12-09T16:52:40.7822888Z target_dir D:\a\deno\deno\target\release
2019-12-09T16:52:40.7823084Z target_dir D:\a\deno\deno\target\release
2019-12-09T16:52:40.7823290Z stdout Hello from plugin. data: test | zero_copy: test
2019-12-09T16:52:40.7823496Z Plugin Sync Response: test
2019-12-09T16:52:40.7824000Z 
2019-12-09T16:52:40.7824961Z stderr 
2019-12-09T16:52:40.7825174Z 
2019-12-09T16:52:40.7825352Z #
2019-12-09T16:52:40.7825549Z # Fatal error in , line 0
2019-12-09T16:52:40.7825745Z # ignored
2019-12-09T16:52:40.7825933Z #
2019-12-09T16:52:40.7826118Z #
2019-12-09T16:52:40.7826282Z #
2019-12-09T16:52:40.7826478Z #FailureMessage Object: 0000008B0EDFD330
2019-12-09T16:52:40.7826848Z ==== C stack trace ===============================
2019-12-09T16:52:40.7827031Z 
2019-12-09T16:52:40.7827264Z 	v8::base::debug::StackTrace::StackTrace [0x00007FF6C68D499B+27]
2019-12-09T16:52:40.7827511Z 	v8::platform::DefaultPlatform::GetStackTracePrinter [0x00007FF6C6726317+55]
2019-12-09T16:52:40.7827724Z 	V8_Fatal [0x00007FF6C6724DB9+217]
2019-12-09T16:52:40.7827936Z 	v8::internal::GlobalBackingStoreRegistry::Register [0x00007FF6C6832689+441]
2019-12-09T16:52:40.7828137Z 	v8::ArrayBuffer::GetBackingStore [0x00007FF6C6715636+262]
2019-12-09T16:52:40.7828350Z 	deno::Send [0x00007FF6C672110C+284]
2019-12-09T16:52:40.7828564Z 	v8::internal::FunctionCallbackArguments::Call [0x00007FF6C6AE557D+589]
2019-12-09T16:52:40.7828784Z 	v8::internal::Builtins::InvokeApiFunction [0x00007FF6C6AE49CA+2378]
2019-12-09T16:52:40.7829031Z 	v8::internal::Builtin_HandleApiCall [0x00007FF6C6AE3FFC+1020]
2019-12-09T16:52:40.7829248Z 	v8::internal::Builtin_HandleApiCall [0x00007FF6C6AE3C47+71]
2019-12-09T16:52:40.7829447Z 	Builtins_CEntry_Return1_DontSaveFPRegs_ArgvOnStack_BuiltinExit [0x00007FF6C695559D+61]
2019-12-09T16:52:40.7829671Z 	Builtins_InterpreterEntryTrampoline [0x00007FF6C68E125B+187]
2019-12-09T16:52:40.7829888Z 	Builtins_InterpreterEntryTrampoline [0x00007FF6C68E125B+187]
2019-12-09T16:52:40.7830093Z 	Builtins_InterpreterEntryTrampoline [0x00007FF6C68E125B+187]
2019-12-09T16:52:40.7830297Z 	Builtins_InterpreterEntryTrampoline [0x00007FF6C68E125B+187]
2019-12-09T16:52:40.7830502Z 	Builtins_InterpreterEntryTrampoline [0x00007FF6C68E125B+187]
2019-12-09T16:52:40.7830685Z 	Builtins_GeneratorPrototypeNext [0x00007FF6C6913072+146]
2019-12-09T16:52:40.7830888Z 	Builtins_JSEntryTrampoline [0x00007FF6C68DEADE+94]
2019-12-09T16:52:40.7831220Z 	Builtins_JSEntry [0x00007FF6C68DE6CC+204]
2019-12-09T16:52:40.7831476Z 	v8::internal::Execution::Call [0x00007FF6C67C7AF6+1062]
2019-12-09T16:52:40.7831688Z 	v8::internal::Execution::Call [0x00007FF6C67C77C3+243]
2019-12-09T16:52:40.7831891Z 	v8::internal::SourceTextModule::ExecuteModule [0x00007FF6C67CFA3F+287]
2019-12-09T16:52:40.7832117Z 	v8::internal::SourceTextModule::InnerModuleEvaluation [0x00007FF6C67CF27F+1071]
2019-12-09T16:52:40.7832338Z 	v8::internal::SourceTextModule::Evaluate [0x00007FF6C67CEDB2+146]
2019-12-09T16:52:40.7832567Z 	v8::internal::SourceTextModule::EvaluateMaybeAsync [0x00007FF6C67CEB5E+302]
2019-12-09T16:52:40.7832785Z 	v8::internal::Module::Evaluate [0x00007FF6C67CA905+101]
2019-12-09T16:52:40.7832990Z 	v8::Module::Evaluate [0x00007FF6C6711226+486]
2019-12-09T16:52:40.7833169Z 	deno_mod_evaluate [0x00007FF6C670C206+230]
2019-12-09T16:52:40.7833381Z 	ZN4deno7isolate7Isolate12mod_evaluate17hcc2f0fdc8c14dcb6E [0x00007FF6C662899E+30]
2019-12-09T16:52:40.7833621Z 	ZN80_$LT$std..future..GenFuture$LT$T$GT$$u20$as$u20$core..future..future..Future$GT$4poll17h4a15c2b98a574942E [0x00007FF6C625DCDB+1963]
2019-12-09T16:52:40.7833877Z 	ZN80_$LT$std..future..GenFuture$LT$T$GT$$u20$as$u20$core..future..future..Future$GT$4poll17hb38b650467cbede1E [0x00007FF6C625FE54+1044]
2019-12-09T16:52:40.7834127Z 	ZN97_$LT$futures_util..compat..compat03as01..Compat$LT$Fut$GT$$u20$as$u20$futures..future..Future$GT$4poll17h4d9b699aa4581661E [0x00007FF6C61A4F33+83]
2019-12-09T16:52:40.7834500Z 	ZN7futures6future5chain22Chain$LT$A$C$B$C$C$GT$4poll17hd679173e7149e6ecE [0x00007FF6C610EF11+97]
2019-12-09T16:52:40.7834722Z 	ZN7futures9task_impl3std3set17hc40a385e2375be91E [0x00007FF6C6602280+96]
2019-12-09T16:52:40.7834924Z 	ZN3std9panicking3try7do_call17hcb4923d650ed8541E.llvm.5887183077164218732 [0x00007FF6C6601F35+149]
2019-12-09T16:52:40.7835175Z 	panic_unwind::__rust_maybe_catch_panic [0x00007FF6C66F4272+34] (/rustc/4560ea788cb760f0a34127156c78e2552949f734\/src\libpanic_unwind\lib.rs:80)
2019-12-09T16:52:40.7835400Z 	ZN16tokio_threadpool4task4Task3run17hb52befe50050fc40E [0x00007FF6C660B8B5+245]
2019-12-09T16:52:40.7835623Z 	ZN16tokio_threadpool6worker6Worker3run17h62be210ba81dc782E [0x00007FF6C6600136+3366]
2019-12-09T16:52:40.7835913Z 	ZN16tokio_threadpool6worker6Worker3run17h62be210ba81dc782E [0x00007FF6C65FF813+1027]
2019-12-09T16:52:40.7836620Z 	ZN3std6thread5local17LocalKey$LT$T$GT$4with17h602f66a0b54c5d82E [0x00007FF6C65E10D9+153]
2019-12-09T16:52:40.7836991Z error: test failed, to rerun pass '-p test_plugin --test integration_tests'
2019-12-09T16:52:40.7837134Z 	ZN3std6thread5local17LocalKey$LT$T$GT$4with17h952ea0e63369e2c8E [0x00007FF6C65E1354+212]
2019-12-09T16:52:40.7837397Z 	ZN96_$LT$tokio_threadpool..park..boxed..BoxedPark$LT$T$GT$$u20$as$u20$tokio_executor..park..Park$GT$6unpark17ha9dd4ebec68c4809E [0x00007FF6C65E87E0+224]
2019-12-09T16:52:40.7837642Z 	ZN3std6thread5local17LocalKey$LT$T$GT$4with17h31781bbcad31b5fbE [0x00007FF6C6607CD3+147]
2019-12-09T16:52:40.7837871Z 	ZN3std6thread5local17LocalKey$LT$T$GT$4with17h9dd6dab994fa9906E [0x00007FF6C6608257+151]
2019-12-09T16:52:40.7838103Z 	ZN3std10sys_common9backtrace28__rust_begin_short_backtrace17h1bca0a7c71f16f26E [0x00007FF6C660AD21+241]
2019-12-09T16:52:40.7838317Z 	ZN3std9panicking3try7do_call17hc8e4e3ad30f41392E.llvm.5887183077164218732 [0x00007FF6C6601E8F+31]
2019-12-09T16:52:40.7838698Z 	panic_unwind::__rust_maybe_catch_panic [0x00007FF6C66F4272+34] (/rustc/4560ea788cb760f0a34127156c78e2552949f734\/src\libpanic_unwind\lib.rs:80)
2019-12-09T16:52:40.7838941Z 	ZN3std6thread7Builder5spawn17he4f870f7a6333807E [0x00007FF6C6602B90+992]
2019-12-09T16:52:40.7839191Z 	alloc::boxed::{{impl}}::call_once<(),FnOnce<()>> [0x00007FF6C66D0817+87] (/rustc/4560ea788cb760f0a34127156c78e2552949f734\src\liballoc\boxed.rs:922)
2019-12-09T16:52:40.7839448Z 	std::sys::windows::thread::{{impl}}::new::thread_start [0x00007FF6C66F22F7+119] (/rustc/4560ea788cb760f0a34127156c78e2552949f734\/src\libstd\sys\windows\thread.rs:47)
2019-12-09T16:52:40.7839671Z 	BaseThreadInitThunk [0x00007FFA68647974+20]
2019-12-09T16:52:40.7839872Z 	RtlUserThreadStart [0x00007FFA6B3AA271+33]
2019-12-09T16:52:40.7840014Z 
2019-12-09T16:52:40.7840238Z thread 'basic' panicked at 'assertion failed: output.status.success()', test_plugin\tests\integration_tests.rs:36:3
2019-12-09T16:52:40.7840478Z note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace.
2019-12-09T16:52:40.7840639Z 
2019-12-09T16:52:40.7840810Z 
2019-12-09T16:52:40.7840986Z failures:
2019-12-09T16:52:40.7841181Z     basic
2019-12-09T16:52:40.7841340Z 
2019-12-09T16:52:40.7841528Z test result: FAILED. 0 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out
2019-12-09T16:52:40.7841702Z 
2019-12-09T16:52:41.0053770Z ##[error]Process completed with exit code 1.
2019-12-09T16:52:41.0106046Z Cleaning up orphan processes
2019-12-09T16:52:41.0892995Z Terminate orphan process: pid (3560) (sccache)

Please add

// Copyright 2018-2019 the Deno authors. All rights reserved. MIT license.

Yeah - we can switch to tar for all platforms. I agree it’s probably better.

Here is an attempted solution which doesn't quite work.

/// For example, to allow automated conversion from Local<String> to
/// Local<Value>
impl<'sc, T, U> From<Local<'sc, T>> for Local<'sc, U>
where
  T: Deref<Target = U>,
{
  fn from(v: Local<'sc, T>) -> Local<'sc, U> {
    unsafe { std::mem::transmute(v) }
  }
}

In C++ this is done by this bit of code: https://github.com/v8/v8/blob/63da8397bc967232f91fea07b1861a50d20ca717/include/v8.h#L192-L201

cc @paulmillr

closed in favor of #47

Let's just revert all the changes I made to .gn

They were arrived through much trial and error and I was hoping it could just be simplified - but apparently not. We can try to simplify in the future. It's really not so important right now.

The problem is only with the windows crate... as a short term solution we could remove support for windows crate and publish 0.26