[ruby-core:102831] [Ruby master Bug#17719] Irregular evaluation order in hash literals

[ruby-core:102831] [Ruby master Bug#17719] Irregular evaluation order in hash literals

[nobu]

Issue #17719 has been reported by nobu (Nobuyoshi Nakada).

----------------------------------------
Bug #17719: Irregular evaluation order in hash literals
https://bugs.ruby-lang.org/issues/17719

* Author: nobu (Nobuyoshi Nakada)
* Status: Open
* Priority: Normal
* Backport: 2.5: UNKNOWN, 2.6: UNKNOWN, 2.7: UNKNOWN, 3.0: UNKNOWN
----------------------------------------
@mame pointed out an irregular evaluation order example.

```sh
$ ruby -e '{foo:p(1), bar:p(2), foo:p(3)}'
-e:1: warning: key :foo is duplicated and overwritten on line 1
1
3
2
```

It feels like a bug.
https://github.com/nobu/ruby/tree/duplicated-keys-order

Or, probably it would be better to turn into an error?



-- 
https://bugs.ruby-lang.org/

[ruby-core:102836] [Ruby master Bug#17719] Irregular evaluation order in hash literals

[xtkoba+ruby]

Issue #17719 has been updated by xtkoba (Tee KOBAYASHI).


To me the evaluation order of 1->3->2 is a bit unexpected, but it might be OK unless specified otherwise.

IMO, duplicate keys in hash literals should be allowed in a scripting language like Ruby. For example, they are explicitly allowed in another scripting language named Python [1]. On the other hand, compiled languages should disallow them.

[1] https://mail.python.org/pipermail/python-ideas/2019-March/055726.html

----------------------------------------
Bug #17719: Irregular evaluation order in hash literals
https://bugs.ruby-lang.org/issues/17719#change-90894

* Author: nobu (Nobuyoshi Nakada)
* Status: Open
* Priority: Normal
* Backport: 2.5: UNKNOWN, 2.6: UNKNOWN, 2.7: UNKNOWN, 3.0: UNKNOWN
----------------------------------------
@mame pointed out an irregular evaluation order example.

```sh
$ ruby -e '{foo:p(1), bar:p(2), foo:p(3)}'
-e:1: warning: key :foo is duplicated and overwritten on line 1
1
3
2
```

It feels like a bug.
https://github.com/nobu/ruby/tree/duplicated-keys-order

Or, probably it would be better to turn into an error?



-- 
https://bugs.ruby-lang.org/

[ruby-core:104611] [Ruby master Bug#17719] Irregular evaluation order in hash literals

[nobu]

Issue #17719 has been updated by nobu (Nobuyoshi Nakada).


I think it can be fixed by removing the optimization in the parser and leaving it to the compiler.

----------------------------------------
Bug #17719: Irregular evaluation order in hash literals
https://bugs.ruby-lang.org/issues/17719#change-92898

* Author: nobu (Nobuyoshi Nakada)
* Status: Open
* Priority: Normal
* Backport: 2.5: UNKNOWN, 2.6: UNKNOWN, 2.7: UNKNOWN, 3.0: UNKNOWN
----------------------------------------
@mame pointed out an irregular evaluation order example.

```sh
$ ruby -e '{foo:p(1), bar:p(2), foo:p(3)}'
-e:1: warning: key :foo is duplicated and overwritten on line 1
1
3
2
```

It feels like a bug.
https://github.com/nobu/ruby/tree/duplicated-keys-order

Or, probably it would be better to turn into an error?



-- 
https://bugs.ruby-lang.org/

[ruby-core:105636] [Ruby master Bug#17719] Irregular evaluation order in hash literals

[jeremyevans0 (Jeremy Evans)]

Issue #17719 has been updated by jeremyevans0 (Jeremy Evans).


nobu (Nobuyoshi Nakada) wrote in #note-2:
> I think it can be fixed by removing the optimization in the parser and leaving it to the compiler.

I tried the approach of removing the optimization from the parser completely (https://github.com/jeremyevans/ruby/commit/bead10831e4f7ea6fc517fe66796f12579915fb3), but it doesn't handle duplicate keys in keyword arguments, resulting in incorrect unknown keyword errors: https://github.com/jeremyevans/ruby/runs/3900240357#step:15:168

I also tried rebasing your `duplicated-keys-order` branch against master, and that appears to fix the issue. I submitted a pull request for that: https://github.com/ruby/ruby/pull/4969

----------------------------------------
Bug #17719: Irregular evaluation order in hash literals
https://bugs.ruby-lang.org/issues/17719#change-94126

* Author: nobu (Nobuyoshi Nakada)
* Status: Open
* Priority: Normal
* Backport: 2.5: UNKNOWN, 2.6: UNKNOWN, 2.7: UNKNOWN, 3.0: UNKNOWN
----------------------------------------
@mame pointed out an irregular evaluation order example.

```sh
$ ruby -e '{foo:p(1), bar:p(2), foo:p(3)}'
-e:1: warning: key :foo is duplicated and overwritten on line 1
1
3
2
```

It feels like a bug.
https://github.com/nobu/ruby/tree/duplicated-keys-order

Or, probably it would be better to turn into an error?



-- 
https://bugs.ruby-lang.org/