Implement const fn {size,align}_of. #42859

eddyb · 2017-06-23T14:52:09Z

nikomatsakis

One nit, one question I just wanted to clarify.

nikomatsakis · 2017-06-27T19:35:25Z

src/libcore/mem.rs

+#[inline]
+#[stable(feature = "rust1", since = "1.0.0")]
+#[cfg(not(stage0))]
+pub const fn size_of<T>() -> usize {


Not that I necessarily object to this, but this would be an "insta-stable" change, right?

This change itself yes, but calling any function in a const context still requires #[feature(const_fn)].

Oh, does it? I thought we enabled calls, but not declarations, of const fns.

Or at least I couldn't remember which it is.

There’s discussion of doing this, but it hasn’t happened yet as far as #24111 says.

nikomatsakis · 2017-06-27T19:36:59Z

src/librustc_mir/transform/qualify_consts.rs

+                                is_shuffle = true;
+                            }
+
+                            _ => {}


maybe add assert!(!self.tcx.is_const_fn(def_id))?

nikomatsakis · 2017-06-27T19:38:06Z

This is pretty simple. =)

eddyb · 2017-06-27T19:48:30Z

@nikomatsakis Yeah, this could've been done months ago...

nikomatsakis · 2017-06-27T21:16:40Z

on IRC, @eddyb noted that we ought to have some tests. Indeed!

I'm basically happy here. @rust-lang/lang -- everybody happy with making size_of and align_of "const fns"?

nikomatsakis · 2017-06-27T21:17:43Z

r=me with tests, presuming that it still takes a feature gate to call a const fn

whitequark · 2017-06-27T21:27:22Z

Can we unreserve sizeof and alignof now? :D Presumably offsetof actually has to be a keyword, at least I don't see an elegant way to implement it with const fns alone. Could be a builtin macro though.

eddyb · 2017-06-27T21:34:10Z

@whitequark I would wait until you can use these functions from constants in stable.
I'm surprised we even bothered to reserve them! I only really want a keyword for typeof 😇.

whitequark · 2017-06-27T21:38:06Z

I only really want a keyword for typeof .

What about offsetof? The uses are pretty marginal but when you need them, you do need them...

eddyb · 2017-06-27T21:47:19Z

@whitequark I'd just use a macro for that. Or have some sort of static reflection capabilities that allow e.g. "indexing by a field" to access the value, safely etc.

retep998 · 2017-06-27T21:53:24Z

I really want offsetof because it would make working with unsized structs for FFI so much easier.

whitequark · 2017-06-27T22:04:00Z

I really want offsetof because it would make working with unsized structs for FFI so much easier.

I think it shouldn't be a lot of work to implement... I wish I had more time to work on rustc.

bors · 2017-06-28T03:44:20Z

☔ The latest upstream changes (presumably #42417) made this pull request unmergeable. Please resolve the merge conflicts.

alevy · 2017-06-28T20:18:26Z

This rocks so hard!

aturon · 2017-07-03T15:52:11Z

@nikomatsakis sounds great to me!

durka · 2017-07-05T04:52:15Z

[00:10:31] error[E0061]: this function takes 1 parameter but 0 parameters were supplied
[00:10:31]    --> /checkout/src/librustc_const_eval/eval.rs:344:34
[00:10:31]     |
[00:10:31] 344 |           if tcx.type_of(def_id).fn_sig().abi() == Abi::RustIntrinsic {
[00:10:31]     |                                  ^^^^^^ expected 1 parameter
[00:10:31]
[00:10:32] error: aborting due to previous error(s)
[00:10:32]
[00:10:32] error: Could not compile `rustc_const_eval`.

nikomatsakis · 2017-07-06T21:27:04Z

@bors r+

bors · 2017-07-06T21:27:05Z

📌 Commit 123799d has been approved by nikomatsakis

eddyb · 2017-07-06T22:03:30Z

@nikomatsakis There are still no testcases FWIW.

nikomatsakis · 2017-07-17T10:13:53Z

Still waiting on tests. =)

eddyb · 2017-07-19T11:49:05Z

@bors r=nikomatsakis

bors · 2017-07-19T11:49:06Z

📌 Commit 148718b has been approved by nikomatsakis

bors · 2017-07-19T11:49:15Z

⌛ Testing commit 148718b with merge 42caa5d8031350c501bba1fb32b7ff8b93892474...

bors · 2017-07-19T11:52:16Z

💔 Test failed - status-travis

retep998 · 2017-07-19T11:58:08Z

Macs are having some issues on travis

The command "curl -fo /usr/local/bin/sccache https://s3.amazonaws.com/rust-lang-ci/rust-ci-mirror/2017-05-12-sccache-x86_64-apple-darwin" failed 3 times.

The command "brew update" failed 3 times.

kennytm · 2017-07-19T12:21:23Z

@retep998 https://www.traviscistatus.com/incidents/74lc35kpxs0y

Due to a regression, some OSX builds have briefly been routed to Linux images. This has change been reverted. We are closely monitoring the situation. Restarting a affected jobs should run the build on the right infrastructure. We apologize for the inconveniences caused.

What's wrong with all the CI problems today… 😕

Mark-Simulacrum · 2017-07-19T12:26:39Z

@bors retry

bors · 2017-07-19T12:26:48Z

⌛ Testing commit 148718b with merge 4ada3a1d0887c79cfbf293fe46d2e7fee18a20ab...

bors · 2017-07-19T12:30:48Z

💔 Test failed - status-appveyor

retep998 · 2017-07-19T12:33:00Z

I wonder which revocation server is offline and who is responsible for keeping it online?

  [35] SSL connect error (schannel: next InitializeSecurityContext failed: Unknown error (0x80092013) - The revocation function was unable to check revocation because the revocation server was offline.)

kennytm · 2017-07-19T12:36:00Z

@retep998 Probably related https://appveyor.statuspage.io/incidents/m2vdvw39kdk8

The issue is a "broken" route between Rackspace data center and US-based CDN edge server for api.nuget.org. Unfortunately, there is no ETA yet. We continue working with nuget team to resolve the issue as soon as possible.

alexcrichton · 2017-07-19T14:12:04Z

@bors: retry

…

On Wed, Jul 19, 2017 at 7:36 AM, kennytm ***@***.***> wrote: @retep998 <https://github.com/retep998> Probably related https://appveyor.statuspage.io/incidents/m2vdvw39kdk8 The issue is a "broken" route between Rackspace data center and US-based CDN edge server for api.nuget.org. Unfortunately, there is no ETA yet. We continue working with nuget team to resolve the issue as soon as possible. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#42859 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAD95IA6tLAJQ2VF5zisqDTO26V5BFK2ks5sPfhDgaJpZM4ODqxq> .

Mark-Simulacrum · 2017-07-19T14:18:43Z

@bors retry -- appveyor revocation certs, appears to be fixed once #43333 lands.

bors · 2017-07-19T16:58:07Z

⌛ Testing commit 148718b with merge 9bbbd29...

@nikomatsakis

Implement const fn {size,align}_of. Fixes #34078. r? @nikomatsakis

bors · 2017-07-19T20:12:48Z

☀️ Test successful - status-appveyor, status-travis
Approved by: nikomatsakis
Pushing 9bbbd29 to master...

rust-highfive assigned nikomatsakis Jun 23, 2017

petrochenkov added the relnotes Marks issues that should be documented in the release notes of the next release. label Jun 23, 2017

shepmaster added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jun 23, 2017

frewsxcv mentioned this pull request Jun 23, 2017

mem::size_of() is not a const function #34078

Closed

frewsxcv added the T-libs-api Relevant to the library API team, which will review and decide on the PR/issue. label Jun 23, 2017

eddyb requested a review from nikomatsakis June 23, 2017 17:46

nikomatsakis requested changes Jun 27, 2017

View reviewed changes

eddyb force-pushed the const-size-and-align-of branch from be029e8 to 7615c86 Compare July 1, 2017 06:37

SimonSapin mentioned this pull request Jul 2, 2017

Individualize feature gates for const fn invocation #43017

Merged

eddyb force-pushed the const-size-and-align-of branch 3 times, most recently from 3600acf to 123799d Compare July 5, 2017 10:55

nikomatsakis approved these changes Jul 6, 2017

View reviewed changes

scottmcm mentioned this pull request Jul 17, 2017

Add align_offset intrinsic and [T]::align_to function rust-lang/rfcs#2043

Merged

Implement const fn {size,align}_of.

148718b

eddyb force-pushed the const-size-and-align-of branch from 1e58b2d to 148718b Compare July 19, 2017 11:47

bors added a commit that referenced this pull request Jul 19, 2017

Auto merge of #42859 - eddyb:const-size-and-align-of, r=nikomatsakis

9bbbd29

Implement const fn {size,align}_of. Fixes #34078. r? @nikomatsakis

bors merged commit 148718b into rust-lang:master Jul 19, 2017

eddyb deleted the const-size-and-align-of branch July 19, 2017 21:03

kennytm mentioned this pull request Jul 20, 2017

size_of of associated type ICEs #43357

Closed

scottmcm mentioned this pull request Jul 31, 2017

Array lengths don't support generic parameters. #43408

Open

scottmcm mentioned this pull request Jan 19, 2018

Issues are not feature requests rust-lang/rfcs#2299

Merged

Badel2 mentioned this pull request Jan 30, 2018

Turn type_id into a constant intrinsic #47892

Merged

aidanhs mentioned this pull request Feb 4, 2018

Allow more const fns in std rust-lang/rfcs#1835

Closed

SimonSapin mentioned this pull request Apr 11, 2018

Add to_bytes and from_bytes to primitive integers #49871

Merged

Implement const fn {size,align}_of. #42859

Implement const fn {size,align}_of. #42859

Conversation

eddyb commented Jun 23, 2017

nikomatsakis left a comment

Choose a reason for hiding this comment

nikomatsakis Jun 27, 2017

Choose a reason for hiding this comment

SimonSapin Jun 27, 2017

Choose a reason for hiding this comment

nikomatsakis Jun 27, 2017

Choose a reason for hiding this comment

nikomatsakis Jun 27, 2017

Choose a reason for hiding this comment

SimonSapin Jun 27, 2017

Choose a reason for hiding this comment

durka Jul 3, 2017

Choose a reason for hiding this comment

nikomatsakis Jun 27, 2017

Choose a reason for hiding this comment

nikomatsakis commented Jun 27, 2017

eddyb commented Jun 27, 2017

nikomatsakis commented Jun 27, 2017

nikomatsakis commented Jun 27, 2017

whitequark commented Jun 27, 2017

eddyb commented Jun 27, 2017

whitequark commented Jun 27, 2017

eddyb commented Jun 27, 2017

retep998 commented Jun 27, 2017

whitequark commented Jun 27, 2017

bors commented Jun 28, 2017

alevy commented Jun 28, 2017

aturon commented Jul 3, 2017

durka commented Jul 5, 2017

nikomatsakis commented Jul 6, 2017

bors commented Jul 6, 2017

eddyb commented Jul 6, 2017

nikomatsakis commented Jul 17, 2017

eddyb commented Jul 19, 2017

bors commented Jul 19, 2017

bors commented Jul 19, 2017

bors commented Jul 19, 2017

retep998 commented Jul 19, 2017

kennytm commented Jul 19, 2017 • edited Loading

Mark-Simulacrum commented Jul 19, 2017

bors commented Jul 19, 2017

bors commented Jul 19, 2017

retep998 commented Jul 19, 2017

kennytm commented Jul 19, 2017

alexcrichton commented Jul 19, 2017 via email

Mark-Simulacrum commented Jul 19, 2017

bors commented Jul 19, 2017

bors commented Jul 19, 2017

kennytm commented Jul 19, 2017 •

edited

Loading