src: do proper StringBytes error handling #12765

addaleax · 2017-04-30T17:30:02Z

Return MaybeLocals from StringBytes::Encode
Add an error out parameter to pass JS exceptions to the callers (instead of directly throwing)
Simplify some of the string generation methods in string_bytes.cc by unifying the EXTERN_APEX logic
Reduce usage of deprecated V8 APIs.
Remove error handling logic from JS, the buffer.*Slice() methods now throw errors themselves.
Left TODO comments for future semver-major error message improvements.

This paves the way for better error messages coming out of the StringBytes methods.

Ref: #3175

Checklist

make -j4 test (UNIX), or vcbuild test (Windows) passes
commit message follows commit guidelines

Affected core subsystem(s)

buffer, fs, os, crypto

CI: https://ci.nodejs.org/job/node-test-commit/9528/

Maybe @TimothyGu or @tniessen would want to review?

Also, /cc @MylesBorins , if you’re still interested in #3175 … that should become a lot easier after this :)

benjamingr · 2017-04-30T17:33:03Z

lib/buffer.js

@@ -590,11 +589,8 @@ Buffer.prototype.toString = function(encoding, start, end) {

    if (end <= start)


This is much clearer now, had to double check undefined <= undefined returns false due to number hint.

@benjamingr Not sure, what are you saying? 😄 I didn’t really touch this part of the logic ;)

You added an early return which eliminates a correct (but very tricky) behavior in the old code where it would check if end <= start for undefined values. I'm just saying I'm happy with the change :)

benjamingr · 2017-04-30T17:34:31Z

src/node_crypto.cc

  delete[] md_value;
-  args.GetReturnValue().Set(rc);
+  // TODO(addaleax): Throw `error` here. This isn't so terribly important,


So why not throw the error here?

Hm, right, I could just do that here. I didn’t want to accidentally turn this into a semver-major change, but I guess here it really doesn’t make a difference… Updated!

benjamingr · 2017-04-30T17:37:23Z

src/string_bytes.cc

+MaybeLocal<Value> StringBytes::Encode(Isolate* isolate,
+                                      const uint16_t* buf,
+                                      size_t buflen,
+                                      Local<Value>* error) {


Personally, I would use a reference and not a pointer here - passing Isolate as a pointer makes sense since it's convention - but if the goal is to literally a const pointer to an object that might change (but not the pointer) I think a reference is cleaner. If this is convention and we can't do a reference - can we make it a const pointer?

(goes to all sites)

Personally, I agree. But yes, we don’t use non-const references in this codebase, and I can live with that… ¯\_(ツ)_/¯

If this is convention - can we make it a const pointer?

Sorry, which thing should be made const? The Local can’t be const, Encode() needs to be able to assign to it. The pointer itself could be const, but it’s a pass-by-value function parameter, so that doesn’t really matter anyway, does it?

The pointer itself, so it's clear it doesn't change - but I don't feel strongly about this and it's plenty obvious it won't change (like the isolate).

benjamingr · 2017-04-30T17:42:16Z

src/string_bytes.cc


    case BASE64: {
      size_t dlen = base64_encoded_size(buflen);
      char* dst = node::UncheckedMalloc(dlen);
      if (dst == nullptr) {
-        return Local<String>();
+        goto malloc_failed;


I think a function containing the logic of malloc_failed and call sites calling the functions (rather than using goto is cleaner. I don't have a lot of experience with this code though.

I’ve just inlined the malloc_failed case, it was a lot bulkier and made more sense this way when I started editing. 😄

benjamingr

I have several nits:

~~Lots of new TODOs that can probably be fixed in this PR pretty quickly.~~
~~I think a reference is cleaner (or at least a const pointer)~~
~~Not a goto fan.~~

That said, I gave the code a read, understand what it does and the changes look good to me with or without the ~~nits addressed - given we can always address them later~~ nits addressed.

I'd feel more comfortable if someone more experienced with the code gave this a review too though :)

addaleax · 2017-04-30T17:46:48Z

Lots of new TODOs that can probably be fixed in this PR pretty quickly.

I’d like to keep them for a quick semver-major follow-up PR after this lands, if that’s okay.

benjamingr · 2017-04-30T17:47:54Z

@addaleax sounds good to me. Editing previous comment.

cjihrig · 2017-05-01T15:06:29Z

lib/buffer.js

  if (arguments.length === 0) {
-    result = this.utf8Slice(0, this.length);
+    return this.utf8Slice(0, this.length);


By changing this to return, you can drop the else and save a level of indentation on the rest of the code below.

@cjihrig yup, done!

- Return `MaybeLocal`s from `StringBytes::Encode` - Add an `error` out parameter to pass JS exceptions to the callers (instead of directly throwing) - Simplify some of the string generation methods in `string_bytes.cc` by unifying the `EXTERN_APEX` logic - Reduce usage of deprecated V8 APIs. - Remove error handling logic from JS, the `buffer.*Slice()` methods now throw errors themselves. - Left TODO comments for future semver-major error message improvements. This paves the way for better error messages coming out of the StringBytes methods. Ref: nodejs#3175

tniessen · 2017-05-01T18:45:05Z

Nothing to add, looks good :)

addaleax · 2017-05-03T13:55:03Z

Fresh CI: https://ci.nodejs.org/job/node-test-commit/9604/

addaleax · 2017-05-03T17:23:33Z

Landed in d56a7e6

- Return `MaybeLocal`s from `StringBytes::Encode` - Add an `error` out parameter to pass JS exceptions to the callers (instead of directly throwing) - Simplify some of the string generation methods in `string_bytes.cc` by unifying the `EXTERN_APEX` logic - Reduce usage of deprecated V8 APIs. - Remove error handling logic from JS, the `buffer.*Slice()` methods now throw errors themselves. - Left TODO comments for future semver-major error message improvements. This paves the way for better error messages coming out of the StringBytes methods. Ref: #3175 PR-URL: #12765 Reviewed-By: Benjamin Gruenbaum <[email protected]> Reviewed-By: James M Snell <[email protected]> Reviewed-By: Tobias Nießen <[email protected]>

- Return `MaybeLocal`s from `StringBytes::Encode` - Add an `error` out parameter to pass JS exceptions to the callers (instead of directly throwing) - Simplify some of the string generation methods in `string_bytes.cc` by unifying the `EXTERN_APEX` logic - Reduce usage of deprecated V8 APIs. - Remove error handling logic from JS, the `buffer.*Slice()` methods now throw errors themselves. - Left TODO comments for future semver-major error message improvements. This paves the way for better error messages coming out of the StringBytes methods. Ref: nodejs#3175 PR-URL: nodejs#12765 Reviewed-By: Benjamin Gruenbaum <[email protected]> Reviewed-By: James M Snell <[email protected]> Reviewed-By: Tobias Nießen <[email protected]>

gibfahn · 2017-06-18T23:22:04Z

Should this be backported to v6.x-staging? If yes please follow the guide and raise a backport PR, if no let me know or add the dont-land-on label.

nodejs-github-bot added c++ Issues and PRs that require attention from people who are familiar with C++. lib / src Issues and PRs related to general changes in the lib or src directory. labels Apr 30, 2017

benjamingr reviewed Apr 30, 2017

View reviewed changes

benjamingr approved these changes Apr 30, 2017

View reviewed changes

cjihrig reviewed May 1, 2017

View reviewed changes

jasnell approved these changes May 1, 2017

View reviewed changes

addaleax added 4 commits May 1, 2017 19:03

[squash] throw errors for crypto return

2ea112b

[squash] remove goto → inline error statements

a65a5c3

[squash] remove indentation for the fast case

389af5a

addaleax force-pushed the stringbytes-cleanup branch from dc00cac to 389af5a Compare May 1, 2017 17:04

addaleax closed this May 3, 2017

addaleax deleted the stringbytes-cleanup branch May 3, 2017 17:23

jasnell mentioned this pull request May 11, 2017

8.0.0 Release Proposal #12220

Closed

addaleax mentioned this pull request Jun 2, 2017

fs.readFileSync can't return a string for a big file #9489

Closed

gibfahn mentioned this pull request Jun 15, 2017

Auditing for 6.11.1 nodejs/Release#230

Closed

3 tasks

gibfahn added the backport-requested-v6.x label Jun 18, 2017

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

src: do proper StringBytes error handling #12765

src: do proper StringBytes error handling #12765

addaleax commented Apr 30, 2017 •

edited

Loading

benjamingr Apr 30, 2017

addaleax Apr 30, 2017

benjamingr Apr 30, 2017

benjamingr Apr 30, 2017

addaleax Apr 30, 2017

benjamingr Apr 30, 2017 •

edited

Loading

addaleax Apr 30, 2017

benjamingr Apr 30, 2017

benjamingr Apr 30, 2017

addaleax Apr 30, 2017

benjamingr left a comment •

edited

Loading

addaleax commented Apr 30, 2017

benjamingr commented Apr 30, 2017

cjihrig May 1, 2017

addaleax May 1, 2017

tniessen commented May 1, 2017

addaleax commented May 3, 2017

addaleax commented May 3, 2017

gibfahn commented Jun 18, 2017

		@@ -590,11 +589,8 @@ Buffer.prototype.toString = function(encoding, start, end) {

		if (end <= start)

src: do proper StringBytes error handling #12765

src: do proper StringBytes error handling #12765

Conversation

addaleax commented Apr 30, 2017 • edited Loading

Checklist

Affected core subsystem(s)

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

benjamingr Apr 30, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

benjamingr left a comment • edited Loading

Choose a reason for hiding this comment

addaleax commented Apr 30, 2017

benjamingr commented Apr 30, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

tniessen commented May 1, 2017

addaleax commented May 3, 2017

addaleax commented May 3, 2017

gibfahn commented Jun 18, 2017

addaleax commented Apr 30, 2017 •

edited

Loading

benjamingr Apr 30, 2017 •

edited

Loading

benjamingr left a comment •

edited

Loading