Harmony 2.8.0 #1509

alexlamsl · 2017-02-26T20:42:42Z

Tests aren't passing at the moment, as this is just a "straight-forward" merge of source files.

(#1508 was created against the wrong branch. Sorry about that.)

…that it is not a reserved word. This should apply to propsmangle as well.

…fiers verify that property names after mangle are legal

closes mishoo#1437

remove extraneous call to AST_SymbolRef.reference() closes mishoo#1443

they are computed but never used closes mishoo#1444

happens when inner function: - just below top level - not referenced - `unused` is disabled closes mishoo#1445

move out of unsafe, guard corner case with screw_id8 instead closes mishoo#1446

closes mishoo#1449

- assign statement does not count towards variable usage by default - only works with assignments on the same scope level as declaration - can be disabled with `unused` set to "keep_assign" - `toplevel` to drop unused top-level variables and/or functions - `top_retain` to whitelist top-level exceptions closes mishoo#1450

fixes - [a].join() => "" + a - ["a", , "b"].join() => "a,,b" - ["a", null, "b"].join() => "a,,b" - ["a", undefined, "b"].join() => "a,,b" improvements - ["a", "b"].join(null) => "anullb" - ["a", "b"].join(undefined) => "a,b" - [a + "b", c].join("") => a + "b" + c closes mishoo#1453

- `do{...}while(false)` => `{...}` - clean up `AST_While` logic closes mishoo#1452

N = 2: a; b; c; d; was: a, b; c; d; now: a, b; c, d; fixes mishoo#1455 closes mishoo#1457

shuffle associative operations to minimise parentheses and aid other uglification efforts closes mishoo#1454

- remove extra tree scanning phase for `negate_iife` - `negate_iife` now only deals with the narrowest form, i.e. IIFE sitting directly under `AST_SimpleStatement` - `booleans`, `conditionals` etc. will now take care the rest via more accurate accounting - `a(); void b();` => `a(); b();` fixes mishoo#1288 closes mishoo#1451

closes mishoo#1459

previously test cases with the same name would be skipped except for the last one `test/run-test.js` will now report duplicated names as errors closes mishoo#1461

- utilise in_use_ids instead of unreferenced() - drop_unused now up-to-date for subsequent passes closes mishoo#1476

- update modified flag between compress() passes - support IIFE arguments - fix corner case with multiple definitions closes mishoo#1473

fix invalid boolean conversion now exposed in `make_node_from_constant()` closes mishoo#1477

- support arrays, objects & AST_Node - support `"a.b":1` on both cli & API - emit warning if variable is modified - override top-level variables fixes mishoo#1416 closes mishoo#1198 closes mishoo#1469

- `test/benchmark.js` measures performance - `test/jetstream.js` verifies correctness - configurable mangle/compress/output options closes mishoo#1479

reduce whitespaces from if-else statements fixes mishoo#1482 closes mishoo#1483

closes mishoo#1487

fixes mishoo#1377 closes mishoo#1488

- `Array.prototype.slice` => `[].slice` closes mishoo#1491

- never exceed specified limit - otherwise warning is shown - enabled only for final output closes mishoo#1496

- only drops side-effect-free arguments - drop side-effect-free parts with discarded value from `AST_Seq` & `AST_SimpleStatement` closes mishoo#1494

fixes mishoo#1492 closes mishoo#1493

A function call or IIFE with an immediately preceding comment containing `@__PURE__` or `#__PURE__` is deemed to be a side-effect-free pure function call and can potentially be dropped. Depends on `side_effects` option. `[#@]__PURE__` hint will be removed from comment when pure call is dropped. fixes mishoo#1261 closes mishoo#1448

@const

- fix corner cases in `const` optimisation - deprecate `/*@const*/` fixes mishoo#1497 closes mishoo#1498

kzc · 2017-02-27T07:11:37Z

master:

$ echo 'var a = /*#__PURE__*/(function(){return 1})();' | bin/uglifyjs -c
var a=function(){return 1}();

#1509 including patch:

$ echo 'var a = /*#__PURE__*/(function(){return 1})();' | bin/uglifyjs -c
WARN: Dropping __PURE__ call [-:1,21]
var a=function(){return 1}();

Ideas?

kzc · 2017-02-27T07:22:50Z

The real question is why didn't master also report the false positive?

alexlamsl · 2017-02-27T07:37:27Z

@kzc I did the commit all on my mobile phone, so I'm afraid I can't debug any further atm.

Let's me get back to you in a couple of hours when I'm back at my desk.

kzc · 2017-02-27T08:00:02Z

On master has_pure_annotation() is not called for:

$ echo 'var a = /*#__PURE__*/(function(){return 1})();' | bin/uglifyjs -c --comments=all
var a=/*#__PURE__*/function(){return 1}();

whereas this harmony PR does call it:

$ echo 'var a = /*#__PURE__*/(function(){return 1})();' | bin/uglifyjs -c --comments=all 
WARN: Dropping __PURE__ call [-:1,21]
var a=/* */function(){return 1}();

so your original concern about false positive warnings was well founded.

I know this is a cop out, but we can drop the warning altogether on both master and harmony. The output is still correct on harmony. It just appears that this harmony PR indirectly calls has_pure_annotation() for whatever reason - thus triggering the false positive warning, while master does not.

Edit:

Note: in spite of the #__PURE__ comment being dropped in the false positive warning case, AST_Call.pure would still be set to true, so the compress code will function correctly.

alexlamsl · 2017-02-27T08:41:51Z

@kzc sounds like a reasonable workaround. I'll give it some more thoughts while getting stuck on the metro.

alexlamsl · 2017-02-27T08:53:26Z

The one thing that made me feel uneasy is the indiscriminate elimination of @__PURE__ in the comments, regardless of whether the IIFE is dropped.

I can cleanly report the warning within drop_side_effect_free(), so one plan of attack is to convert existing sites which checks for has_side_effects() for the purpose of wholesome elimination to use the aforementioned method. Now I wish I can get home sooner!

- consolidate `side-effects` optimisations - improve string `+` optimisation - enhance literal & `conditionals` optimisations

fixes mishoo#1510

alexlamsl · 2017-02-27T18:49:13Z

@kzc a0eaff7 corresponds to #1510 (comment)

alexlamsl · 2017-02-27T19:30:08Z

#1493 now exposes a bug with AST_ObjectProperty.key.start.pos on harmony

update exception messages

alexlamsl · 2017-02-27T20:11:56Z

The only other test failure left is to do with -source-map-inline giving slightly different content

`master`

{
  "version":3,
  "sources":[
    "test/input/issue-1323/sample.js"
  ],
  "names":[
    "bar",
    "foo"
  ],
  "mappings":"AAAA,GAAIA,KAAO,WACP,QAASC,KAAKD,KACV,MAAOA,KAGX,MAAOC"
}

This PR

{
  "version":3,
  "sources":[
    "test/input/issue-1323/sample.js"
  ],
  "names":[
    "bar",
    "foo"
  ],
  "mappings":"AAAA,GAAIA,KAAM,WACN,QAASC,KAAKD,KACV,MAAOA,KAGX,MAAOC"
}

kzc · 2017-02-27T21:12:30Z

#1493 now exposes a bug with AST_ObjectProperty.key.start.pos on harmony

That's minor. Can the test simply be if(0)'d out on the harmony branch with a comment?

kzc · 2017-02-27T21:20:02Z

Regarding the ---source-map-inline failure, has the output changed due to compress improvements and/or differences?

kzc · 2017-02-27T22:20:56Z

@alexlamsl Looks like you've made really good progress. The remaining test failures are related to wrong start/end positions of AST_Nodes it appears. I guess there were harmony parser changes that inadvertently dropped this info.

alexlamsl · 2017-02-28T05:22:42Z

Ah, so it was broken before, but b7bb706 then broke the test output instead of fixing the parser.

Restoring sourceMappingURL to the same as master and the test now passes.

not sure if `start.pos` is correct, but oh well

alexlamsl · 2017-02-28T06:10:16Z

All tests passed now 💯

alexlamsl · 2017-02-28T06:13:25Z

@avdg @kzc PTAL

kzc · 2017-02-28T15:01:25Z

@alexlamsl Congrats on the 2.8.0 releases.

kzc · 2017-02-28T15:11:22Z

I assumed this PR was merged already. :-)

Looks good to merge with a harmony-v2.8.0 tag.

avdg · 2017-03-02T13:57:01Z

Too much to review on my side, sorry :P

alexlamsl · 2017-03-02T14:17:21Z

@avdg no worries ;-)

Anat Dagan and others added 30 commits February 10, 2017 14:13

in mangle_names there is a check that the variable name is legal and …

81f1df1

…that it is not a reserved word. This should apply to propsmangle as well.

Merge pull request mishoo#1481 from anatdagan/propsmangle_only_identi…

eb55d8a

…fiers verify that property names after mangle are legal

Improve optimizing function() { if(c){return foo} bar();}

dd31d12

closes mishoo#1437

fix crash in unsafe replacement of undefined

11676f9

remove extraneous call to AST_SymbolRef.reference() closes mishoo#1443

remove unused AST_Scope.nesting & AST_SymbolRef.frame

686a496

they are computed but never used closes mishoo#1444

fix corner case in keep_fnames

fa668a2

happens when inner function: - just below top level - not referenced - `unused` is disabled closes mishoo#1445

enable typeof "undefined" for general use

e5badb9

move out of unsafe, guard corner case with screw_id8 instead closes mishoo#1446

fix stray else in compress with conditionals=false

d11dca3

closes mishoo#1449

tweak do-while loops

ae4db00

- `do{...}while(false)` => `{...}` - clean up `AST_While` logic closes mishoo#1452

-c sequences=N suboptimal at N expression cutoff

f584ca8

N = 2: a; b; c; d; was: a, b; c; d; now: a, b; c, d; fixes mishoo#1455 closes mishoo#1457

improve string concatenation

6b3c49e

shuffle associative operations to minimise parentheses and aid other uglification efforts closes mishoo#1454

smarter const replacement taking name length into account

6ffbecb

closes mishoo#1459

fix duplicated test names

c525a2b

previously test cases with the same name would be skipped except for the last one `test/run-test.js` will now report duplicated names as errors closes mishoo#1461

improve keep_fargs & keep_fnames

b8b133d

- utilise in_use_ids instead of unreferenced() - drop_unused now up-to-date for subsequent passes closes mishoo#1476

improve reduce_vars and fix a bug

a0f4fd3

- update modified flag between compress() passes - support IIFE arguments - fix corner case with multiple definitions closes mishoo#1473

evaluate AST_SymbolRef as parameter

974247c

fix invalid boolean conversion now exposed in `make_node_from_constant()` closes mishoo#1477

enhance global_defs

e275148

- support arrays, objects & AST_Node - support `"a.b":1` on both cli & API - emit warning if variable is modified - override top-level variables fixes mishoo#1416 closes mishoo#1198 closes mishoo#1469

add benchmark & JetStream tests

7e6331b

- `test/benchmark.js` measures performance - `test/jetstream.js` verifies correctness - configurable mangle/compress/output options closes mishoo#1479

improve --beautify bracketize

09f9ae2

reduce whitespaces from if-else statements fixes mishoo#1482 closes mishoo#1483

Add .gitattributes to checkout lf eol style

c06a50f

closes mishoo#1487

remove extraneous spaces between ++/+/--/-

ac0b61e

fixes mishoo#1377 closes mishoo#1488

introduce unsafe_proto

ec64acd

- `Array.prototype.slice` => `[].slice` closes mishoo#1491

clean up max_line_len

8898b8a

- never exceed specified limit - otherwise warning is shown - enabled only for final output closes mishoo#1496

fix pure_funcs & improve side_effects

26fbeec

- only drops side-effect-free arguments - drop side-effect-free parts with discarded value from `AST_Seq` & `AST_SimpleStatement` closes mishoo#1494

Fix: AST_Accessor missing start / end tokens

d48a308

fixes mishoo#1492 closes mishoo#1493

enable collapse_vars & reduce_vars by default

4e49302

- fix corner cases in `const` optimisation - deprecate `/*@const*/` fixes mishoo#1497 closes mishoo#1498

alexlamsl mentioned this pull request Feb 27, 2017

warn & drop #__PURE__ iff IIFE is dropped #1511

Merged

alexlamsl and others added 3 commits February 28, 2017 02:25

warn & drop #__PURE__ iff IIFE is dropped (mishoo#1511)

858e6c7

- consolidate `side-effects` optimisations - improve string `+` optimisation - enhance literal & `conditionals` optimisations

Merge branch 'master' into harmony-2.8.0

57777b6

fix parsing of arrow function with bind

a0eaff7

fixes mishoo#1510

alexlamsl mentioned this pull request Feb 27, 2017

[ES6]Failure to parse arrow function with bind #1510

Closed

fix parser tests

a942dc0

update exception messages

alexlamsl added 2 commits February 28, 2017 13:23

restore sourceMappingURL in test

ab21753

fix parser test

478aaab

not sure if `start.pos` is correct, but oh well

alexlamsl changed the title ~~[WIP] Harmony 2.8.0~~ Harmony 2.8.0 Feb 28, 2017

alexlamsl merged commit 514fc68 into mishoo:harmony Feb 28, 2017

alexlamsl deleted the harmony-2.8.0 branch February 28, 2017 15:14

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Harmony 2.8.0 #1509

Harmony 2.8.0 #1509

alexlamsl commented Feb 26, 2017

kzc commented Feb 27, 2017

kzc commented Feb 27, 2017

alexlamsl commented Feb 27, 2017

kzc commented Feb 27, 2017 •

edited

Loading

alexlamsl commented Feb 27, 2017

alexlamsl commented Feb 27, 2017

alexlamsl commented Feb 27, 2017

alexlamsl commented Feb 27, 2017

alexlamsl commented Feb 27, 2017

kzc commented Feb 27, 2017

kzc commented Feb 27, 2017

kzc commented Feb 27, 2017 •

edited

Loading

alexlamsl commented Feb 28, 2017

alexlamsl commented Feb 28, 2017

alexlamsl commented Feb 28, 2017

kzc commented Feb 28, 2017

kzc commented Feb 28, 2017

avdg commented Mar 2, 2017

alexlamsl commented Mar 2, 2017

Harmony 2.8.0 #1509

Harmony 2.8.0 #1509

Conversation

alexlamsl commented Feb 26, 2017

kzc commented Feb 27, 2017

kzc commented Feb 27, 2017

alexlamsl commented Feb 27, 2017

kzc commented Feb 27, 2017 • edited Loading

alexlamsl commented Feb 27, 2017

alexlamsl commented Feb 27, 2017

alexlamsl commented Feb 27, 2017

alexlamsl commented Feb 27, 2017

alexlamsl commented Feb 27, 2017

master

This PR

kzc commented Feb 27, 2017

kzc commented Feb 27, 2017

kzc commented Feb 27, 2017 • edited Loading

alexlamsl commented Feb 28, 2017

alexlamsl commented Feb 28, 2017

alexlamsl commented Feb 28, 2017

kzc commented Feb 28, 2017

kzc commented Feb 28, 2017

avdg commented Mar 2, 2017

alexlamsl commented Mar 2, 2017

kzc commented Feb 27, 2017 •

edited

Loading

`master`

kzc commented Feb 27, 2017 •

edited

Loading