Bug #12376: No error for elseif condition

• Status: Rejected
• Assignee: -
• Target version: -
• ruby -v: 2.3.0
• Backport: 2.1: UNKNOWN, 2.2: UNKNOWN, 2.3: UNKNOWN

[dsferreira (Daniel Ferreira)]

Something I have been for a long time wondering about.
I don't know if this is a parsing problem or not but for me it seems like it is.

Replication:

$ irb
> RUBY_VERSION
 => '2.3.0'
> a = 1
> if a == 2
>   puts 2
> elsif a == 3
>   puts 3
> elseif a == 1
>   puts 1
> else
>   puts 0
> end
0
 => nil

Shouldn't we have a parse error triggered here?
We can replace elseif for any other instruction like foo for instance.

I believe this behaviour is around for a long time.

[dsferreira (Daniel Ferreira)]

• ruby -v changed from 2.30 to 2.3.0

[zverok (Victor Shepelev)]

It is parsed as

if a == 2
  puts 2
elsif a == 3
  puts 3
  elseif(a == 1)
  puts 1
else
  puts 4
end

With elseif being just a method call. With this branch not evaluated, there is no problem with unknown method in it.

[matz (Yukihiro Matsumoto)]

• Status changed from Open to Closed

it's parsed like

a = 1
if a == 2
  puts 2
  elseif(a==1)
  puts 1
else
  puts 0
end

and the method elseif is not called in the false condition.
But you can find this kind of bugs using rubocop or other tools

Matz.

[nobu (Nobuyoshi Nakada)]

And most editors can high-light keywords, but not elseif, as shown in this page, FYI.

[usa (Usaku NAKAMURA)]

• Status changed from Closed to Rejected

[dsferreira (Daniel Ferreira)]

Of course.
Makes perfect sense.
I would say the way to handle this bugs in a pro-active way would be to use 100% coverage in the test suite.
I can not see how the bug would be missed with that.
TDD as a must in ruby like always.
Thank you for the enlightenment Sirs.

[mame (Yusuke Endoh)]

You may want to use a code lint tool, such as rubocop.

BTW, next quiz:

> a = 1
> if a == 2
>   puts 2;
> elsif a == 3
>   puts 3.
> elsif a == 1
>   puts 1;
> else
>   puts 0;
> end
0
 => nil

--
Yusuke Endoh [e-mail]