Collections Reform Part 2 #509

Gankra · 2014-12-09T17:54:21Z

This RFC shores up the finer details of collections reform. In particular, where the previous RFC
focused on general conventions and patterns, this RFC focuses on specific APIs. It also patches
up any errors that were found during implementation of part 1. Some of these changes
have already been implemented, and simply need to be ratified.

rendered

Gankra · 2014-12-09T17:56:34Z

CC @aturon @reem @huonw @cmr @pcwalton @nodakai @bfops

One of your desired collections items may have arrived in this RFC.

arielb1 · 2014-12-09T18:20:11Z

We can do append by just having a Vec::<T>::clear_into_iter::<'a>(&'a mut self) -> MoveItems<T>+'a that sets len to 0. If we want to hack performance we can do it by branching on TypeId, or equivalently on the return value of Any – see http://is.gd/6bznhA (LLVM optimizes this out).

ftxqxd · 2014-12-09T18:54:01Z

text/0000-collections-reform-part-2.md

+```
+/// Gets a slice over all the elements in the RingBuf. This may require shifting
+/// all the elements to make this possible.
+pub fn to_slice(&'a self) -> &'a [T]


Wouldn’t this require &mut self to allow shifting the elements inside?

aliblong · 2014-12-09T20:23:29Z

I prefer reserve_index for VecMap at least. In the example given in the documentation, where a VecMap maps month names to numbers, my thought process would be "How many months are there? 12! So I want to reserve 12 in this VecMap". It would feel clumsy to have to reserve 13, especially given that 0-indexing has taught my brain to think in n-1 with collections.

It's also important to note that whatever design we choose should be mirrored in with_capacity.

sfackler · 2014-12-09T20:25:44Z

text/0000-collections-reform-part-2.md

+
+* Stabilize `front`/`back`/`front_mut`/`back_mut` for peeking on the ends of Deques
+
+* Explicitly specify HashMap's iterators to be non-deterministic between iterations. This would


It seems like this is hacking around the fact that there are types that want to implement ExactSizeIterator but not DoubleEndedIterator. Why don't we clean up that API?

Most things that care about ExactSize also care about DoubleEnded, so I'm not sure if that's totally helpful. I agree that it's a borderline abuse to impl DoubleEnded on an "unordered" sequence, though.

aturon · 2014-12-11T05:52:53Z

Passing thought: should with_capacity on e.g. VecMap be renamed as well?

aliblong · 2014-12-11T06:09:29Z

Perhaps rename with_capacity to with_reserved_index for VecMap? A bit verbose, but fits nicely with reserve_index.

tbu- · 2014-12-11T10:09:36Z

With the PR rust#19715 I raised the question whether VecMap (and consequently Bitv, TrieMap as @gankro pointed out) should have a type parameter for the key type. As this lets us add more possible index types in a backward-compatible fashion, I'm very much for it. Could that be added to this RFC?

nodakai · 2014-12-12T04:09:41Z

text/0000-collections-reform-part-2.md

+based around the largest index they have set, and not the number of elements they contain.
+One could instead opt for `reserve_len` or
+`reserve_capacity`, which are effectively the same method, but with an off-by-one. That is,
+`reserve_len(x)` == reserve_index(x - 1)`. `reserve_len` would have the benefit of returning the


Just a nitpicking; markup error here.

`reserve_len(x)` == reserve_index(x - 1)`

Gankra · 2014-12-12T22:01:04Z

@aturon And I sat down this afternoon and played a little marry/murder/make-out with libcollections. As a result major update:

RFC now recommends stabilization/removal paths for all collections
RFC now deprecates some Vec and DList methods
split_at -> split_off
reserve_len wins
-> (&[T], &[T]) wins for RingBuf::as_slices

alexcrichton · 2014-12-12T23:07:44Z

text/0000-collections-reform-part-2.md

+///
+/// Calls either `grow()` or `truncate()` depending on whether `new_len`
+/// is larger than the current value of `len()` or not.
+pub fn resize(&mut self, new_len: uint, value: T) where T: Clone


Out of curiosity, could you explain a little more the rationale for including this convenience method? I'm not sure I've personally ever needed it, but I may just be in a different line of work!

This is something a few people has asked for. I don't have a strong desire/motivation for it, but I thought I would give it a fair try out in this RFC.

Asked for in rust-lang/rust#19104 (comment)

so maybe @nodakai can give a better justification.

Ah ok, I may be forgetting, but do we have a number of other convenience methods like this on collections? I thought we had pared it back to a fairly bare-bones sense.

If one wants to ensure that a vector is exactly 10 items long, this is the best method to call since it will work right even if the vector is currently 9 or 11 items long.

IMO grow and truncate are the convenience methods, not resize.

It would be worth looking at what the C++ std::vector provides: http://en.cppreference.com/w/cpp/container/vector

(Spoiler: It's only resize.)

@Valloric But growing and shrinking really are very different operations. resize necessarily must branch on whether to grow or shrink, whereas grow and truncate don't.

@Valloric it would be difficult to remove truncate as you can't always manifest a T to call resize to truncate. C++ looks like it leverages overloading as well to provide some more functionality which we may not be able to match exactly.

alexcrichton · 2014-12-12T23:17:07Z

This looks awesome @gankro, nice work!

Gankra · 2014-12-13T23:38:43Z

@thestinger: for the sake of transparency, I feel obligated to notify you that the current draft of this RFC is recommending the removal of Tree*, Trie*, LruCache, bitflags!, and EnumSet from the standard library. I believe you were opposed to this in the past.

bluss · 2014-12-15T10:41:11Z

text/0000-collections-reform-part-2.md

+* Unstable: Bitv, BitvSet, VecMap
+* Move to [collect-rs](https://github.com/Gankro/collect-rs/) for incubation:
+EnumSet, bitflags!, LruCache, TreeMap, TreeSet, TrieMap, TrieSet
+


Do you think DList is ready? It's unclear if it is near its final form.

And why do you think bitflags isn't ready? I agree about other collections.

Collections Reform 1 already proposed cutting out a lot of DList's methods, so it's going to have a pretty minimal/uncontroversial interface.

As for bitflags, I'll leave that to @aturon

Add drain, and link to WIP implementations.

Gankra · 2014-12-17T04:16:27Z

Added a drain iterator for moving out of an &mut Collection without wasting the allocation.

This pull request: *Renames `BinaryHeap::top` to `BinaryHeap::peek` *Stabilizes `front/back/front_mut/back_mut` in `DList` and `RingBuf` *Stabilizes `swap` in `RingBuf` in accordance with rust-lang/rfcs#509. Note that this PR does not address `Bitv::{get,set}` or HashMap's iterators, nor does it move `std::vec` to `std::collections::vec`, all of which still need to be done. Because of the method renaming, this is a [breaking-change].

bluss · 2014-12-24T23:02:44Z

What did you intend w.r.t ExactSizeIterator and DoubleEndedIterator? I'm against implementing traits just because it's possible, with the hash maps and sets it doesn't seem useful at all. Maybe ExactSizeIterator on its own if it is untangled from DEI, but even there I don't really know any real use.

Implement `reserve_len` and `reserve_len_exact` for `VecMap` in accordance with rust-lang/rfcs#509.

This PR deprecates the `DList::ListInsertion` trait, in accordance with rust-lang/rfcs#509. The functions which were previously part of the ListInsertion impl for `DList::IterMut` have been moved to be inherent methods on the iterator itself, and appropriate doctests have been added.

This commit takes a second pass through the `vec` module to stabilize its API. The changes are as follows: **Stable**: * `dedup` * `from_raw_parts` * `insert` * `into_iter` * `is_empty` * `remove` * `reserve_exact` * `reserve` * `retain` * `swap_remove` * `truncate` **Deprecated**: * `from_fn`, `from_elem`, `grow_fn` and `grow`, all deprecated in favor of iterators. See rust-lang/rfcs#509 * `partition`, `partitioned`, deprecated in favor of a new, more general iterator consumer called `partition`. * `unzip`, deprecated in favor of a new, more general iterator consumer called `unzip`. A few remaining methods are left at experimental status. [breaking-change]

bluss · 2015-01-02T16:46:40Z

Is it possible to amend the RFC to keep .merge() on DList? It seems absurd to add new "allocation-free" method to DList while removing others like .merge(). I'm mostly confused by the cluelessness of the deprecation message "Use other methods on IterMut"; it would be better to be more honest and tell the user there is no replacement.

Gankra · 2015-01-02T17:00:56Z

I assure you merge was never allocation free; it just used IterMut.

Gankra · 2015-01-02T17:07:04Z

Oh no wait, it totally did! I misread the source.

Implements the `append()` and `split_off()` methods proposed by the collections reform part 2 RFC. RFC: rust-lang/rfcs#509 Tracking issue: #19986

sandeep-datta · 2015-03-30T12:21:15Z

I was told to leave a comment here instead of opening a new issue:
Rename string::push_str to string::concat #20857

Please consider renaming string::push_str() to string::concat(). Push_X() in the context of a collection means X will be added to the container as a single entity. Hence push_str() is meaningless for strings since a string can only contain characters (char != [char]).

If it is too late to rename then consider adding a method called concat() anyway.

Also consider renaming Vec::push_all() to Vec::concat() so that the name is consistent with String::concat().

ftxqxd reviewed Dec 9, 2014
View reviewed changes

sfackler reviewed Dec 9, 2014
View reviewed changes

aturon mentioned this pull request Dec 11, 2014

Make collections::VecMap generic over its key type rust-lang/rust#19715

Closed

brson assigned aturon Dec 11, 2014

nodakai reviewed Dec 12, 2014
View reviewed changes

alexcrichton reviewed Dec 12, 2014
View reviewed changes

aturon mentioned this pull request Dec 13, 2014

Stabilization metabug: 1.0-alpha rust-lang/rust#19260

Closed

bluss reviewed Dec 15, 2014
View reviewed changes

cgaebel mentioned this pull request Dec 16, 2014

Add RingBuf::as_slices as per collections reform v2. rust-lang/rust#19903

Merged

Merge pull request #1 from cgaebel/drain

a5a655d

Add drain, and link to WIP implementations.

alexcrichton mentioned this pull request Dec 17, 2014

[hashmap] Adds drain: a way to sneak out the elements while clearing a map. rust-lang/rust#19946

Merged

csouth3 mentioned this pull request Dec 22, 2014

Implement reserve_len and reserve_len_exact for VecMap rust-lang/rust#20143

Merged

csouth3 mentioned this pull request Dec 24, 2014

Deprecate DList::ListInsertion rust-lang/rust#20185

Merged

bors added a commit to rust-lang/rust that referenced this pull request Dec 27, 2014

auto merge of #20143 : csouth3/rust/vecmap-reserve, r=Gankro

18842f8

Implement `reserve_len` and `reserve_len_exact` for `VecMap` in accordance with rust-lang/rfcs#509.

TimDumol mentioned this pull request Jan 1, 2015

Add append() and split_off() to DList as per coll. reform. rust-lang/rust#20406

Merged

drbawb mentioned this pull request Jan 6, 2015

Does not build on latest nightly w/ RFC 509 reem/rust-typemap#21

Closed

erickt mentioned this pull request Jan 17, 2015

Rename string::push_str to string::concat rust-lang/rust#20857

Closed

Gankra mentioned this pull request Jan 19, 2015

add append to vec with tests and benchmarks rust-lang/rust#21330

Merged

aturon mentioned this pull request Aug 12, 2015

Tracking issue for drain stabilization rust-lang/rust#27711

Closed

Gankra mentioned this pull request Oct 23, 2015

Tracking issue for Vec::resize rust-lang/rust#27790

Closed

Centril added the A-collections Proposals about collection APIs label Nov 23, 2018

crlf0710 mentioned this pull request Jan 8, 2020

Tracking issue for RFC 2044: dual-MIT/Apache2 licensing rust-lang/rust#43461

Open

7 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Collections Reform Part 2 #509

Collections Reform Part 2 #509

Gankra commented Dec 9, 2014

Gankra commented Dec 9, 2014

arielb1 commented Dec 9, 2014

ftxqxd Dec 9, 2014

Gankra Dec 9, 2014

aliblong commented Dec 9, 2014

sfackler Dec 9, 2014

Gankra Dec 9, 2014

aturon commented Dec 11, 2014

aliblong commented Dec 11, 2014

tbu- commented Dec 11, 2014

nodakai Dec 12, 2014

Gankra commented Dec 12, 2014

alexcrichton Dec 12, 2014

Gankra Dec 12, 2014

alexcrichton Dec 12, 2014

Valloric Dec 12, 2014

Valloric Dec 12, 2014

Gankra Dec 12, 2014

alexcrichton Dec 12, 2014

alexcrichton commented Dec 12, 2014

Gankra commented Dec 13, 2014

bluss Dec 15, 2014

pczarn Dec 15, 2014

Gankra Dec 15, 2014

Gankra commented Dec 17, 2014

bluss commented Dec 24, 2014

bluss commented Jan 2, 2015

Gankra commented Jan 2, 2015

Gankra commented Jan 2, 2015

sandeep-datta commented Mar 30, 2015


		* Stabilize `front`/`back`/`front_mut`/`back_mut` for peeking on the ends of Deques

		* Explicitly specify HashMap's iterators to be non-deterministic between iterations. This would

Collections Reform Part 2 #509

Collections Reform Part 2 #509

Conversation

Gankra commented Dec 9, 2014

Gankra commented Dec 9, 2014

arielb1 commented Dec 9, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

aliblong commented Dec 9, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

aturon commented Dec 11, 2014

aliblong commented Dec 11, 2014

tbu- commented Dec 11, 2014

Choose a reason for hiding this comment

Gankra commented Dec 12, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

alexcrichton commented Dec 12, 2014

Gankra commented Dec 13, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Gankra commented Dec 17, 2014

bluss commented Dec 24, 2014

bluss commented Jan 2, 2015

Gankra commented Jan 2, 2015

Gankra commented Jan 2, 2015

sandeep-datta commented Mar 30, 2015