Deprecate no_debug and custom_derive #37128

nrc · 2016-10-12T21:12:53Z

nrc · 2016-10-12T21:13:31Z

bors · 2016-10-13T01:33:34Z

☔ The latest upstream changes (presumably #37118) made this pull request unmergeable. Please resolve the merge conflicts.

nikomatsakis

Looks good; left a few suggestions for improvement.

nikomatsakis · 2016-10-17T17:34:02Z

src/librustc_lint/builtin.rs

+                                                           ..) = g {
+                    cx.span_lint(DEPRECATED,
+                                 attr.span,
+                                 &format!("use of deprecated attribute: {}", name));


Nit: probably we want:

"use of deprecated attribute: `{}`" // ^ ^ note ticks

right?

Also, let's link the deprecation to a tracking issue

This follows the format for deprecated items. That probably should have back ticks too (added), but we don't have the tracking issue number anywhere so we can't link to it very easily.

Well for deprecated items we give a reason, right? I imagined adding the issue number (or maybe just a string) to the Deprecated variant of Stability, in any case.

nikomatsakis · 2016-10-17T17:34:36Z

src/librustc_lint/builtin.rs

+        let name = &*attr.name();
+        for &(n, _, ref g) in feature_gate::KNOWN_ATTRIBUTES {
+            if n == name {
+                if let &feature_gate::AttributeGate::Gated(feature_gate::Stability::Deprecated,


Nit: this seems mega-inefficient...but probably it doesn't matter as there aren't that many attributes, and that table ain't that long, I guess?

At minimum we could filter out the list of deprecations once rather than per attribute =)

nikomatsakis · 2016-10-17T17:35:40Z

src/libsyntax/feature_gate.rs

 // fn() is not Debug
 impl ::std::fmt::Debug for AttributeGate {
    fn fmt(&self, fmt: &mut ::std::fmt::Formatter) -> ::std::fmt::Result {
        match *self {
-            Gated(ref name, ref expl, _) => write!(fmt, "Gated({}, {})", name, expl),
+            Gated(_, ref name, ref expl, _) => write!(fmt, "Gated({}, {})", name, expl),


why exclude Stability from the Debug printout?

nrc · 2016-10-18T05:08:55Z

@nikomatsakis changes made

nrc · 2016-10-18T19:55:36Z

More changes for review

bors · 2016-10-19T09:53:38Z

☔ The latest upstream changes (presumably #37269) made this pull request unmergeable. Please resolve the merge conflicts.

nrc · 2016-10-19T18:56:13Z

rebased

nikomatsakis · 2016-10-19T19:51:07Z

tidy errors:

/build/src/libsyntax/feature_gate.rs:672: line longer than 100 chars

nikomatsakis · 2016-10-19T19:51:26Z

r=me once tidy is fixed; thanks

nrc · 2016-10-19T21:36:27Z

@bors: r=nikomatsakis

bors · 2016-10-19T21:36:28Z

📌 Commit f6e0b3a has been approved by nikomatsakis

bors · 2016-10-20T08:09:20Z

⌛ Testing commit f6e0b3a with merge 5694dd1...

bors · 2016-10-20T09:28:24Z

💔 Test failed - auto-win-gnu-32-opt-rustbuild

nikomatsakis · 2016-10-24T21:53:39Z

@nrc

---- [compile-fail] compile-fail\feature-gate-no-debug-2.rs stdout ----

error: C:/bot/slave/auto-win-gnu-32-opt-rustbuild/build/src/test/compile-fail/feature-gate-no-debug-2.rs:14: unexpected "error": '14:1: 14:12: use of deprecated attribute `no_debug`: the `#[no_debug]` attribute is an experimental feature. See https://github.com/rust-lang/rust/issues/29721'

error: C:/bot/slave/auto-win-gnu-32-opt-rustbuild/build/src/test/compile-fail/feature-gate-no-debug-2.rs:14: expected error not found: use of deprecated attribute: no_debug

error: 1 unexpected errors found, 1 expected errors not found
status: exit code: 101
command: PATH="C:\bot\slave\auto-win-gnu-32-opt-rustbuild\build\obj\build\i686-pc-windows-gnu\stage2\bin;C:\bot\slave\auto-win-gnu-32-opt-rustbuild\build\obj\build\i686-pc-windows-gnu\stage2-tools\i686-pc-windows-gnu\release\deps;C:\bot\slave\auto-win-gnu-32-opt-rustbuild\build\obj\build\i686-pc-windows-gnu\stage2-rustc\i686-pc-windows-gnu\release\deps;C:\bot\slave\auto-win-gnu-32-opt-rustbuild\build\obj\build\i686-pc-windows-gnu\stage2-test\i686-pc-windows-gnu\release\deps;C:\bot\slave\auto-win-gnu-32-opt-rustbuild\build\obj\build\i686-pc-windows-gnu\stage2-std\i686-pc-windows-gnu\release\deps;C:\mingw-w64\i686-4.9.1-win32-dwarf-rt_v3-rev1\mingw32\bin;C:\Python27;C:\msys64\mingw32\bin;C:\msys64\usr\bin;C:\Windows\system32;C:\Windows;C:\Windows\System32\Wbem;C:\Windows\System32\WindowsPowerShell\v1.0;C:\msys64\usr\bin;C:\python27;C:\python27\scripts;C:\program files (x86)\inno setup 5;C:\program files (x86)\CMake\bin;C:\Program Files (x86)\Windows Kits\8.1\Windows Performance Toolkit;C:\Program Files\Microsoft SQL Server\110\Tools\Binn;C:\Program Files (x86)\Microsoft SDKs\TypeScript\1.0;C:\Program Files\CMake\bin" C:\bot\slave\auto-win-gnu-32-opt-rustbuild\build\obj\build\i686-pc-windows-gnu\stage2\bin\rustc.exe C:\bot\slave\auto-win-gnu-32-opt-rustbuild\build\src/test\compile-fail\feature-gate-no-debug-2.rs -L C:\bot\slave\auto-win-gnu-32-opt-rustbuild\build\obj\build\i686-pc-windows-gnu\test\compile-fail --target=i686-pc-windows-gnu --error-format json -L C:\bot\slave\auto-win-gnu-32-opt-rustbuild\build\obj\build\i686-pc-windows-gnu\test\compile-fail\feature-gate-no-debug-2.stage2-i686-pc-windows-gnu.compile-fail.libaux -C prefer-dynamic -o C:\bot\slave\auto-win-gnu-32-opt-rustbuild\build\obj\build\i686-pc-windows-gnu\test\compile-fail\feature-gate-no-debug-2.stage2-i686-pc-windows-gnu.exe -Crpath -O -Lnative=C:\bot\slave\auto-win-gnu-32-opt-rustbuild\build\obj\build\i686-pc-windows-gnu\rust-test-helpers
unexpected errors (from JSON output): [
    Error {
        line_num: 14,
        kind: Some(
            Error
        ),
        msg: "14:1: 14:12: use of deprecated attribute `no_debug`: the `#[no_debug]` attribute is an experimental feature. See https://github.com/rust-lang/rust/issues/29721"
    }
]

not found errors (from test file): [
    Error {
        line_num: 14,
        kind: Some(
            Error
        ),
        msg: "use of deprecated attribute: no_debug"
    }
]

thread '[compile-fail] compile-fail\feature-gate-no-debug-2.rs' panicked at 'explicit panic', C:\bot\slave\auto-win-gnu-32-opt-rustbuild\build\src\tools\compiletest\src\runtest.rs:1097
note: Run with `RUST_BACKTRACE=1` for a backtrace.


failures:
    [compile-fail] compile-fail\feature-gate-no-debug-2.rs

nrc · 2016-10-25T01:02:01Z

@bors: r=@nikomatsakis

bors · 2016-10-25T01:02:03Z

📌 Commit 38f993f has been approved by @nikomatsakis

bors · 2016-10-25T04:59:00Z

⌛ Testing commit 38f993f with merge 98cb0c3...

@nikomatsakis

Deprecate no_debug and custom_derive r? @nikomatsakis

bors · 2016-10-25T06:15:59Z

💔 Test failed - auto-win-gnu-32-opt-rustbuild

alexcrichton · 2016-10-25T06:38:20Z

Ok that failure corresponds to a failure to free on Windows. That's terrifying!

@bors: retry

Spurious failure in HeapFree on Windows #37395

bors · 2016-10-25T09:23:45Z

⌛ Testing commit 38f993f with merge 3c4de7f...

bors · 2016-10-25T11:08:17Z

💔 Test failed - auto-mac-64-opt

alexcrichton · 2016-10-25T16:06:26Z

@bors: retry

nikomatsakis · 2016-10-25T17:23:13Z

I'm inclined not to backport at this point. Not because it's harmful per se but just because we can just push deprecation back one cycle, so why not? What do you think @nrc?

nrc · 2016-10-25T17:53:24Z

@nikomatsakis yeah, it is getting kind of late into the cycle for a backport.

nikomatsakis · 2016-10-25T20:05:52Z

Removed beta-nominated tag: decided against backport.

bors · 2016-10-26T09:07:57Z

⌛ Testing commit 38f993f with merge 918bc3a...

bors · 2016-10-26T10:47:54Z

💔 Test failed - auto-linux-64-opt

alexcrichton · 2016-10-26T12:43:34Z

Hm that's two segfaults at the same location on two different platforms, maybe a legitimate segfault?

nrc · 2016-10-26T22:37:39Z

Dammit, Rust is not meant to segfault

Has a custom deprecation since deprecating features is not supported and is a pain to implement

nrc · 2016-10-27T04:29:23Z

@bors: r=@alexcrichton

bors · 2016-10-27T04:29:24Z

📌 Commit 8c4a39c has been approved by @alexcrichton

bors · 2016-10-27T07:51:59Z

⌛ Testing commit 8c4a39c with merge 46d39f3...

@nikomatsakis

Deprecate no_debug and custom_derive r? @nikomatsakis

bors · 2016-10-27T11:06:29Z

nrc · 2016-10-27T19:52:56Z

FTR r was nikomatsakis, not alexcrichton

SergioBenitez · 2016-12-11T06:47:59Z

I'd like to request that this deprecation be reverted, and I plan on submitting a PR to do so. This deprecation penalizes developers that write multiple types of syntax extensions instead of only custom derives, and it seems unnecessary while macros 2.0 are still some time away.

At present, it is possible to have a single codegen crate that implements custom attributes, custom derives, and procedural macros. All of these syntax extensions can be imported into a project with:

#![feature(plugin, custom_derive)]
#![plugin(codegen_crate)]

With this deprecation, however, this is no longer possible.

The alternative, of course, is to use macros 1.1. But custom derives via macros 1.1 must be implemented in a separate crate of type proc-macro. As a result, a second crate must be created. This means that a project using custom derives and custom attributes or procedural macros must implement syntax extensions in two different crates, and users using that project must depend on both crates and import the syntax extensions in two different ways:

#![feature(plugin)]
#![plugin(codegen_crate)]

#[macro_use] extern crate derive_crate;

I believe this will lead to user confusion: syntax extensions are now coming from two different crates, and they must be imported via completely different mechanisms. While before a single declaration brought in all syntax extensions, now two different declarations are used to bring in syntax extensions.

This is also significantly more unergonomic for the developer. This is because it is no longer feasible to easily share code between custom derives and other syntax extensions. The work-arounds are to have one create depend on the other, pushing logic to one "master" crate, or to create a third crate that contains the shared logic. Clearly, neither are as ergonomic as the existing implementation.

All of this issues will likely be ameliorated by the unification macros 2.0 promises to bring. But macros 2.0 won't land in a nightly release any time soon, and they won't be stabilized until quite some time after that. So why deprecate this unnecessarily early? It penalizes developers that write multiple types of syntax extensions. We should keep this around until macros 2.0 are within reach of stabilization.

nrc added the beta-nominated Nominated for backporting to the compiler in the beta channel. label Oct 12, 2016

rust-highfive assigned nikomatsakis Oct 12, 2016

nrc force-pushed the depr-attr branch from dccc671 to fc27992 Compare October 17, 2016 01:49

nikomatsakis requested changes Oct 17, 2016

View reviewed changes

nrc force-pushed the depr-attr branch from 4f7ff39 to a2d08b0 Compare October 18, 2016 19:55

nrc force-pushed the depr-attr branch from a2d08b0 to 1445e81 Compare October 19, 2016 18:55

nikomatsakis approved these changes Oct 19, 2016

View reviewed changes

nrc force-pushed the depr-attr branch from 1445e81 to f6e0b3a Compare October 19, 2016 21:35

alexcrichton added the T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. label Oct 24, 2016

nrc force-pushed the depr-attr branch from f6e0b3a to 38f993f Compare October 25, 2016 01:01

bors added a commit that referenced this pull request Oct 25, 2016

Auto merge of #37128 - nrc:depr-attr, r=@nikomatsakis

98cb0c3

Deprecate no_debug and custom_derive r? @nikomatsakis

nikomatsakis removed the beta-nominated Nominated for backporting to the compiler in the beta channel. label Oct 25, 2016

nrc added 5 commits October 27, 2016 11:44

Add possibility of deprecating attributes

121e903

deprecate no_debug

d1b08b1

Deprecate custom_derive

b7675cf

Has a custom deprecation since deprecating features is not supported and is a pain to implement

deprecation message for custom derive

9322332

review changes

8c4a39c

nrc force-pushed the depr-attr branch from 38f993f to 8c4a39c Compare October 27, 2016 04:28

bors added a commit that referenced this pull request Oct 27, 2016

Auto merge of #37128 - nrc:depr-attr, r=@alexcrichton

46d39f3

Deprecate no_debug and custom_derive r? @nikomatsakis

bors merged commit 8c4a39c into rust-lang:master Oct 27, 2016

SergioBenitez mentioned this pull request Dec 26, 2016

Would a PR to use Macros 1.1 be welcome rwf2/Rocket#50

Closed

SergioBenitez mentioned this pull request Jan 12, 2017

Tracking issue for custom_derive feature #29644

Closed

alexcrichton mentioned this pull request Feb 2, 2017

Tracking issue for no_debug feature #29721

Closed

Deprecate no_debug and custom_derive #37128

Deprecate no_debug and custom_derive #37128

Conversation

nrc commented Oct 12, 2016

nrc commented Oct 12, 2016

bors commented Oct 13, 2016

nikomatsakis left a comment

Choose a reason for hiding this comment

nikomatsakis Oct 17, 2016

Choose a reason for hiding this comment

nikomatsakis Oct 17, 2016

Choose a reason for hiding this comment

nrc Oct 18, 2016

Choose a reason for hiding this comment

nikomatsakis Oct 18, 2016

Choose a reason for hiding this comment

nikomatsakis Oct 17, 2016

Choose a reason for hiding this comment

nikomatsakis Oct 17, 2016

Choose a reason for hiding this comment

nikomatsakis Oct 17, 2016

Choose a reason for hiding this comment

nrc commented Oct 18, 2016

nrc commented Oct 18, 2016

bors commented Oct 19, 2016

nrc commented Oct 19, 2016

nikomatsakis commented Oct 19, 2016

nikomatsakis commented Oct 19, 2016

nrc commented Oct 19, 2016

bors commented Oct 19, 2016

bors commented Oct 20, 2016

bors commented Oct 20, 2016

nikomatsakis commented Oct 24, 2016

nrc commented Oct 25, 2016

bors commented Oct 25, 2016

bors commented Oct 25, 2016

bors commented Oct 25, 2016

alexcrichton commented Oct 25, 2016 • edited Loading

bors commented Oct 25, 2016

bors commented Oct 25, 2016

alexcrichton commented Oct 25, 2016

nikomatsakis commented Oct 25, 2016

nrc commented Oct 25, 2016

nikomatsakis commented Oct 25, 2016

bors commented Oct 26, 2016

bors commented Oct 26, 2016

alexcrichton commented Oct 26, 2016

nrc commented Oct 26, 2016

nrc commented Oct 27, 2016

bors commented Oct 27, 2016

bors commented Oct 27, 2016

bors commented Oct 27, 2016

nrc commented Oct 27, 2016

SergioBenitez commented Dec 11, 2016

alexcrichton commented Oct 25, 2016 •

edited

Loading