From @masak

<diakopter> rakudo​: say 6 >== 2
<p6eval> rakudo c3aaa2​: OUTPUT«Cannot assign to readonly value␤current
instr.​: '&infix​:<=>' [...]
<diakopter> weird.. it's using *any* operator as an assignment
operator, not just infix
<diakopter> I mean, not just non-boolean-returning :)
<diakopter> alpha​: say $_ >== 2
<p6eval> alpha 30e0ed​: OUTPUT«No applicable candidates found to
dispatch to for 'prefix​:=' [...]
<diakopter> std​: say $_ >== 2
<p6eval> std 29742​: OUTPUT«�[31m===�[0mSORRY!�[31m===�[0m␤Can't make
assignment out of >= because chaining operators are diffy [...] FAILED
00​:01 106m␤»
<diakopter> ah, a non-LTA error
* masak submits rakudobug

@coke - Status changed from 'new' to 'open'

From @coke

On Tue Feb 16 02​:04​:32 2010, masak wrote​:

<diakopter> rakudo​: say 6 >== 2
<p6eval> rakudo c3aaa2​: OUTPUT«Cannot assign to readonly value␤current
instr.​: '&infix​:<=>' [...]
<diakopter> weird.. it's using *any* operator as an assignment
operator, not just infix
<diakopter> I mean, not just non-boolean-returning :)
<diakopter> alpha​: say $_ >== 2
<p6eval> alpha 30e0ed​: OUTPUT«No applicable candidates found to
dispatch to for 'prefix​:=' [...]
<diakopter> std​: say $_ >== 2
<p6eval> std 29742​: OUTPUT«�[31m===�[0mSORRY!�[31m===�[0m␤Can't make
assignment out of >= because chaining operators are diffy [...] FAILED
00​:01 106m␤»
<diakopter> ah, a non-LTA error
* masak submits rakudobug

Slight change in behavior (note that Moar behaves differently than parrot here)

02​:42 <[Coke]> rakudo​: say 6 >== 2
02​:43 <camelia> rakudo-jvm e8cea1​: OUTPUT«(timeout)»
02​:43 <camelia> ..rakudo-moar e8cea1​: OUTPUT«No such method 'STORE' for
  invocant of type 'Int'␤ in block at
  src/gen/m-CORE.setting​:16488␤ in block at /tmp/tmpfile​:1␤␤»
02​:43 <camelia> ..rakudo-parrot e8cea1​: OUTPUT«Cannot modify an immutable
  value␤ in block at gen/parrot/CORE.setting​:16739␤ in block
  at /tmp/tmpfile​:1␤␤»

--
Will "Coke" Coleda

From @Mouq

Fixed in Rakudo commits​:

commit d262118eccf5517412c05318ddb50e9a8fefe9e8
Author​: Mouq <alexmoquin@​gmail.com>
Date​: Sat Feb 7 21​:32​:14 2015 -0500

  Add a very general X​::Syntax​::Can'tMeta and use it

commit 34669da476fb9cce4e55ffdc72e89d3dded83cf0
Author​: Mouq <alexmoquin@​gmail.com>
Date​: Sat Feb 7 20​:30​:40 2015 -0500

  Improve metaop​:<=>

  Dies in the case of fiddly/diffy operators, and now adjusts to the
  proper precedence depending on the operator

Behavior now​:

$ perl6 -e'say 6 >== 2'
===SORRY!=== Error while compiling -e
Cannot make assignment out of >= because chaining operators are too diffy
at -e​:1
------> say 6 >==⏏ 2

Tests added in Roast commit​:

commit 54b136e898549abb2d081927cf785ded577d33b0
Author​: Mouq <alexmoquin@​gmail.com>
Date​: Sat Feb 7 21​:48​:01 2015 -0500

  Add test for #​72874

Marking as resolved :)

On Sun Feb 23 23​:44​:08 2014, coke wrote​:

On Tue Feb 16 02​:04​:32 2010, masak wrote​:

<diakopter> rakudo​: say 6 >== 2
<p6eval> rakudo c3aaa2​: OUTPUT«Cannot assign to readonly
value␤current
instr.​: '&infix​:<=>' [...]
<diakopter> weird.. it's using *any* operator as an assignment
operator, not just infix
<diakopter> I mean, not just non-boolean-returning :)
<diakopter> alpha​: say $_ >== 2
<p6eval> alpha 30e0ed​: OUTPUT«No applicable candidates found to
dispatch to for 'prefix​:=' [...]
<diakopter> std​: say $_ >== 2
<p6eval> std 29742​: OUTPUT«�[31m===�[0mSORRY!�[31m===�[0m␤Can't make
assignment out of >= because chaining operators are diffy [...]
FAILED
00​:01 106m␤»
<diakopter> ah, a non-LTA error
* masak submits rakudobug

Slight change in behavior (note that Moar behaves differently than
parrot here)

02​:42 <[Coke]> rakudo​: say 6 >== 2
02​:43 <camelia> rakudo-jvm e8cea1​: OUTPUT«(timeout)»
02​:43 <camelia> ..rakudo-moar e8cea1​: OUTPUT«No such method 'STORE'
for
invocant of type 'Int'␤ in block at
src/gen/m-CORE.setting​:16488␤ in block at
/tmp/tmpfile​:1␤␤»
02​:43 <camelia> ..rakudo-parrot e8cea1​: OUTPUT«Cannot modify an
immutable
value␤ in block at gen/parrot/CORE.setting​:16739␤
in block
at /tmp/tmpfile​:1␤␤»

@Mouq - Status changed from 'open' to 'resolved'