https://issues.dlang.org/show_bug.cgi?id=14232

--- Comment #1 from Kenji Hara <k.hara.pg@gmail.com> ---
The following program already reported same error in 2.066:

struct Bug
{
    const const void bug()
    {
    }
}

test.d(3): Error: redundant attribute 'const'

--

https://issues.dlang.org/show_bug.cgi?id=14232

briancschott@gmail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |briancschott@gmail.com

--- Comment #2 from briancschott@gmail.com ---
Here are my thoughts as somebody who is responsible for ensuring that my employer's D code is ready for new compiler releases: Fixing something like this takes so little time that nobody cares.

Actually, we do care. We care that the compiler helped us fix bad code.

If you want an example of the kind of problem that industry D users face during compiler upgrades, look at #14169 instead of this.

--

https://issues.dlang.org/show_bug.cgi?id=14232

Martin Nowak <code@dawg.eu> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |WONTFIX

--

https://issues.dlang.org/show_bug.cgi?id=14232

Ketmar Dark <ketmar@ketmar.no-ip.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ketmar@ketmar.no-ip.org

--- Comment #3 from Ketmar Dark <ketmar@ketmar.no-ip.org> ---
should any bug be deprecated first — just in case somebody wrote invalid code relying on that bug?

--

https://issues.dlang.org/show_bug.cgi?id=14232

--- Comment #4 from Martin Nowak <code@dawg.eu> ---
(In reply to Ketmar Dark from comment #3)
> should any bug be deprecated first — just in case somebody wrote invalid code relying on that bug?

Usually yes, but within reason, depending on how complicated it is to fix the code compared to how complicated the deprecation is.

--

https://issues.dlang.org/show_bug.cgi?id=14232

--- Comment #5 from Ketmar Dark <ketmar@ketmar.no-ip.org> ---
this effectively turns compiler code to unmaintainable spaghetti. and i believe that this is the perfect application for lint-like tool (yes, devs can finally bless dfix). with dfix blessed and considered the part of the compiler such bigfixes can be accompanied by a PR with new rule for dfix. this will solve the problem and allow keeping frontend code clean.

--

https://issues.dlang.org/show_bug.cgi?id=14232

Dicebot <public@dicebot.lv> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
                 CC|                            |public@dicebot.lv
         Resolution|WONTFIX                     |---

--- Comment #6 from Dicebot <public@dicebot.lv> ---
> Fixing something like this takes so little time that nobody cares.

This is fundamentally flawed attitude I wish was less common in D community. As someone who is responsible for fixing stuff I don't even slightly care if required change is trivial or not - what I care most that the code that used to work (and work correctly!) suddenly stopped compiling with no warning and/or explanation. There shouldn't be legitimate reason for doing things like that ever, not matter how trivial and justified change is.

Breaking changes are awesome but only when they are done right. I want to be sure that if I update the compiler stuff that wasn't warned in previous release will keep compiling. This is bare minimum for planning sane upgrades.

--

https://issues.dlang.org/show_bug.cgi?id=14232

--- Comment #7 from Ketmar Dark <ketmar@ketmar.no-ip.org> ---
so do you mean that if some syntax was allowed by accident, it should be kept forever, shitting compiler code with hacks over hacks trying to accept both valid code and invalid code?

no, really, there is no single reason to accept invalid code. invalid code must not compile. not "silently accepted", not "warned", not "deprecated".

or, taking "POV example", the bare minuimum of sanity in language is not accepting invalid code, regardless of how much somebody wants it to stay with us.

--

https://issues.dlang.org/show_bug.cgi?id=14232

--- Comment #8 from Martin Nowak <code@dawg.eu> ---
(In reply to Ketmar Dark from comment #7)
> so do you mean that if some syntax was allowed by accident, it should be kept forever

No it should be kept for one release with a deprecation warning attached.

--

https://issues.dlang.org/show_bug.cgi?id=14232

--- Comment #9 from Dicebot <public@dicebot.lv> ---
It was introduced in 1b932b9282df6ed312b6da0031417ea0d4f8faa5 I am looking in how deprecation message can be added

--