slides: a sliding window range iterator by wilzbach · Pull Request #4027 · dlang/phobos

wilzbach · 2016-02-27T18:40:23Z

As you can see - I was a bit productive during this weekend ;-)

I propose here to add a pairwise range iterator that iterates over all k-mers of a range, e.g.

[0,1,2].slides(2).array == [[0, 1], [1,2]];

Why is this useful?

For me it's kind of common to work with k-mers and texts and thus it's nice to have a handy iterator for this. See this simple example for how easy counting will get:

int[dstring] e;
"AGAGA"d.slides.each!(a => e[a]++);
assert(e == ["AG"d: 2, "GA"d: 2]);

and of course works like this for any k:

int[dstring] e;
"AGAGA"d.slides(3).each!(a => e[a]++);
assert(e == ["AGA"d: 2, "GAG"d: 1]);

Current alternatives

It's possible to do this with zip, but it looks a bit ugly and is only practical for n==2.

auto arr = "AGAGA".array;
arr.zip(arr.save.dropOne).map!"a.expand.only".map!(to!string).each!(a => d[a]++);

Trivia

I think pairs is a fine name to, but that might conflict with other functions e.g. byPairs
the default is set to 2, as that's the most common use case
the implementation is based on chunks, but I couldn't figure out a proper way to copy/inherit this functionality and most functions are actually changed

edit: changed name to slides

yebblies · 2016-02-28T01:02:47Z

Another way to say it is that it iterates over all sub-strings with a certain length.

wilzbach · 2016-02-29T13:25:44Z

@yebblies you might be right that the documentation and name could be more intuitive. I looked at different standard libs and how they name this function.

tl;dr: slides seem the most intuitive choice.
Alternatively there is also windows or cons - has anyone an opinion one this?

Scala: sliding

An iterator that yields a sliding fixed-sized window of elements in this collection.

sliding

Ruby: each_cons

Iterates the given block for each array of consecutive elements. If no block is given, returns an enumerator.

each_cons

Clojure: partition

Usage: (partition n step pad coll)

Returns a lazy sequence of lists of n items each, at offsets step
apart. If step is not supplied, defaults to n, i.e. the partitions
do not overlap. If a pad collection is supplied, use its elements as
necessary to complete last partition upto n items. In case there are
not enough padding elements, return a partition with less than n items.

partition

quickfur · 2016-02-29T17:36:28Z

The use of "pair" here, both in the function name and in the documentation, is very confusing, because "pair" means two of something, but you're actually talking about k-mers, for any k≥2. (On that note, does the code handle the k=1 case nicely?)

IMO a better name would be something along the lines of a sliding window of size k, or something along those lines.

wilzbach · 2016-02-29T23:30:39Z

The use of "pair" here, both in the function name and in the documentation, is very confusing, because "pair" means two of something, but you're actually talking about k-mers, for any k≥2.
IMO a better name would be something along the lines of a sliding window of size k, or something along those lines.

I renamed it to slides and updated the documentation to go with the term fixed-size sliding window.

On that note, does the code handle the k=1 case nicely?

Yes - check the tests ;-)

burner · 2016-03-02T17:40:51Z

std/range/package.d

+    import std.array: array;
+    import std.algorithm: equal, map;
+    import std.conv: to;
+    assert(iota(3).slides(3).array.to!string == "[[0, 1, 2]]");


there is no reason to convert this to string just compare with [[0,1,2]] or use the algorithm function equal

I didn't know that its as easy as using equal!equal([[0,1,2]])) - very nice :)

D is like vim. You learn something new every day.

wilzbach · 2016-03-04T14:26:46Z

FYI: This is also part of ndslice.
http://dlang.org/phobos/std_experimental_ndslice_selection.html#.windows

wilzbach · 2016-03-04T18:42:49Z

This also tackles parts of issue 6621
https://issues.dlang.org/show_bug.cgi?id=6621

JackStouffer · 2016-03-06T19:43:40Z

std/range/package.d

+        }
+    }
+
+private:


Two nitpicks:

Using Python's notation for private symbols with an underscore is useless with an explicit private feature

Struct members should be at the top

Using Python's notation for private symbols with an underscore is useless with an explicit private feature

I know, but this style keeps popping up in phobos and the method that slides is based on (chunks) used the same notation.

Struct members should be at the top

Again kept the style from chunks, but I do agree that it should be at top - updated :)

wilzbach · 2016-03-08T10:16:23Z

Anything else that is blocking this PR? - I do believe that this is a quite common method and useful for phobos.

JakobOvrum · 2016-03-18T19:41:32Z

std/range/package.d

+of size $(D windowSize) of a $(D source) range.
+$(D Source) must be a forward range. $(D windowSize) must be
+greater than zero. For $(D windowSize = 2) it is similar to
+$D(zip(arr, arr.save.dropOne)).


zip(source, source.save.dropOne)?

edit:

s/$D(/$(D/ (see broken output). Note that we've introduced a new syntax code to replace $(D code) and there shouldn't be any reason to use the D macro directly in new additions.

rcorre · 2016-03-20T20:02:19Z

A bit late to the party, but I've needed this more than once so it would be great to have in phobos.
At the risk of bikeshedding, could it just be called slide? chunks is the only other plural name and it doesn't seem to fit here.

Just for fun, a compile-time windows size gives a succint definition and allows multiple foreach arguments:

http://dpaste.dzfl.pl/fced4f16a987

I haven't figured out a clever way to use that same technique for a runtime windowSize though.

wilzbach · 2016-03-23T23:02:30Z

We generally use nested structs for new higher order ranges, and we generally don't document range primitives individually.
Range primitives always have the same or very similar signature when present, and their presence is often conditional, which is not reflected in the documentation in any special way. I think the signal to noise ratio of higher order range documentation is much better when range capabilities are documented in prose in the main documentation comment.

I fully agree with this, however when I looked at package.d I saw that most of the functions in this file - even similar ones like chunks, evenChunks use a public struct. Are there any plan to move the structs into the function?
(moved the slides inside the function and addressed all comments)

JakobOvrum · 2016-03-23T23:12:51Z

Those are quite old. I'm not sure if they were designed that way because of buggy/incomplete compilers or simply because practice regarding ranges hadn't yet fully evolved. Maybe a mix of both (and maybe there will be someone who comes out of the woodwork in support of the old practice…).

Regardless, changing them would require deprecating the public range types and reworking their documentation, and YMMV whether the latter is enough to justify the former. At any rate I don't know of any plan for them.

wilzbach · 2016-03-25T08:57:01Z

A bit late to the party, but I've needed this more than once so it would be great to have in phobos.
At the risk of bikeshedding, could it just be called slide? chunks is the only other plural name and it doesn't seem to fit here.

ping @andralex - what is your opinion on this addition? slides or slide?

9il · 2016-03-25T09:45:10Z

Please add additional variant, which can work with simple input ranges. This function should be included along with circular buffer range, thought. In addition, two kinds of sliding windows should be supported: one with fixed width and one with max length (less length for borders).

schveiguy · 2016-03-25T14:14:38Z

I think we should not be adding more inner types, because they add significant template bloat. I experienced this to a huge degree when writing my iopipe library.

See this forum post: https://forum.dlang.org/post/n96k3g$ka5$1@digitalmars.com

Note: if it is decided to keep the struct inside the function, there is no need to add the template parameter to the inner struct. It's already templated based on the function.

wilzbach · 2017-02-16T23:56:06Z

Okay I managed to find a bit of time & went through slides again.
In short I addressed the comments, added more tests, cleaned the code a bit more, added explanations for tricky bits (see e.g. opSlice or back) and a changelog entry.

drop @Property :)

Isn't this the typical range style to have empty, front length as @property?
(I know that @property for getter functions is nearly useless, hence "style")

looks like something went too far somewhere with this dollar business... not sure what use cases we're improving

This was inspired by the way chunk handles the dollar tokens.

Consider dropping bidirectional support. The overwhelming use will be forward.
Sliding over input ranges is really a composition of "I need a buffered range that promotes my input range to a range with a lookahead" and, well, the existing code. I think we should leave this PR for the Slider for forward ranges, and then introduce Buffered with a new PR. Then the composition will be straightforward with little or no glue needed.

Alrighty - I removed the buffered sliding window, but left the bidirectional support as it's very short code (once figured out correctly).

JackStouffer

Looks good. One more pass through

JackStouffer · 2017-02-17T15:16:15Z

std/range/package.d

+        enum needsEndTracker = true;
+    }
+
+    private bool _empty;


redundant private

JackStouffer · 2017-02-17T15:16:21Z

std/range/package.d

+
+    static if (hasSlicing!Source)
+    {
+        private enum hasSliceToEnd = hasSlicing!Source && is(typeof(Source.init[0 .. $]) == Source);


JackStouffer · 2017-02-17T15:19:04Z

std/range/package.d

+                    {
+                        // if possible use the faster opDollar overload
+                        static if (hasSliceToEnd)
+                            _source = _source[$ .. $];


Often opDollar is just an alias to the length function. Why do you state that the $ branch is faster?

JackStouffer · 2017-02-17T15:21:12Z

std/range/package.d

+    /// Forward range primitives. Always present.
+    @property auto front()
+    {
+        assert(!empty, "Attempting to access front on an empty range");


nitpick:

Attempting to access front on an empty slides

Same for other error messages

JackStouffer · 2017-02-17T15:24:58Z

std/range/package.d

+    {
+        /**
+        Indexing and slicing operations. Provided only if
+        $(D hasSlicing!Source) is $(D true).


nitpick: backticks

JackStouffer · 2017-02-17T15:25:22Z

std/range/package.d

+
+        static if (!isInfinite!Source)
+        {
+            typeof(this) opSlice(size_t lower, size_t upper)


/// ditto

JackStouffer · 2017-05-05T17:37:40Z

Ping @wilzbach

this needs just a bit more work to be ready, let's get this merged

Comments addressed

JackStouffer · 2017-05-15T13:39:40Z

As @wilzbach seems to be busy I'll pull this and submit a follow up PR to address the nitpicks

slides: a sliding window range iterator merged-on-behalf-of: unknown

Clean-up for PR #4027 merged-on-behalf-of: Andrei Alexandrescu <andralex@users.noreply.github.com>

@disable

* Improve many docs in std.array * Fix missing '|' in std.format grammar. * Add CT-checked format string overloads to std.format * Add overloads for formattedWrite, formattedRead, format, sformat. * Throw FormatError when formatValue is called with floating points. The latter allows `snprintf` to be avoided in CTFE when checking format strings so floating point arguments can be checked. * Tweaks Add constraints. Fix test for ctfpMessage equality. * Add changelog entry * Improve some docs in std.traits * Added some type qualifiers to std.parallelism * Use dfmt on std/concurrency.d * Issue 13568: Add CT-checked format string overloads to std.stdio * Add overloads for readf, writef, writefln. * Separate fmt argument from args for writef[ln] to improve docs. * std.format: Improve FormatException messages * Simplified overloads of std.range.cycle. * std.container.Array - documentation cleanup - add missing details, e.g. about exception being thrown - fix wrong descriptions - fix wrong complexity specifications * changed front declaration to an alias * Document compile-time checking for format strings * split std.algorithm tests to avoid OOM - as seen on win-farm-1 tester * Improve many docs in std.array * Clarify deallocation done by typecons.RefCounted. I had trouble understanding how RefCounted!T was supposed to help me manage resources. These document changes should clarify the process and make it easier for others to understand. This is follow-up after this post on the D learn forum: https://forum.dlang.org/post/qctmkqpqnreysxcjmrgm@forum.dlang.org and after commits cd86cc2 and 33217eb "Clarify deallocation done by std.typecons.Unique." * Fix issue 17270 * std.container.Array - optimize concatenation * fix * assignment and improved unit test * Nicer error messages for std.typecons.experimental.Final. * added static assert for * type * added static asserts for arrays and pointers * Fix issue 17251 - Appender.put doesn't accept const input range elements. The two overloads taking an element and a const range were conflicting because canPutConstRange is overlapping the definition of canPutItem. * Remove trailing whitespace. * Remove @disable, which is now redundant. * Add disclaimer to std.random about unsuitability for cryptographical purposes. * It now still accepts inputRanges when they're infinite. * Add link to mir-random * update information about ndslice * Fixed typos in RefCounted documentation. * More edits to typecons.RefCounted * Mention the extra checks in the std.string.assumeUTF docs * Documented undocumented hmac helpers * Fix Issue 17288 - formattedWrite with width/precision and no value Previously an RangeError at runtime or assert(0) at CT would fire from formatNth with message e.g. "n = 1". * Fix Issue 17286 - A function for comparing two digests securely * Use collectExceptionMsg, fix coverage for missing width/precision * Add tests for incompatible format spec Note: Remaining formatValue overloads may defer to these value types, e.g. pointer -> int, bool -> int, char -> int. * std.container.Array - do not realloc if enough capacity * fix issue 15763 - Document behavior of relative difference. Also fix issue where symmetry was assumed when lhs was a value and rhs was not. * Merge pull request dlang#5296 from ntrel/ct-stdio-format Issue 13568: Add CT-checked format string overloads to std.stdio merged-on-behalf-of: H. S. Teoh <quickfur@users.noreply.github.com> * clear changelog * Added changelog entry for secureCompare * Fix Issue 17283 - std.experimental.typecons uses private module members * Fix issue 15534 - Rework constructor docs, removing reference to string parameter that doesn't exist, and using more appropriate ddoc style. * Add unit test * Make allocators' resolveInternalPointer take a const pointer as source * Allocators should take a ubyte[] memory block instead of a void[] one Allocators that rely on taking a block of unused memory to manage and allocate from it should take a ubyte[] instead of a void[] in order to prevent users from accidentally passing valid data and cause memory corruption/leaks. * Fix style problems in std.experimental.allocator.gc_allocator * Remove now-redundant import of onRangeError. * Changed unit test * update pipeline ci repo url - finally moved repo under dlang * fix issue 17314 * fix issue 17314 * Only grow if full * Fix style * fix link The link is broken because the module name. Hope this will fix it. * Update random.d * Fix issue 17327 - std.getopt: Repeated boolean command option fails. * Fix issue 17327. Review comments: drop continue statement. * netbsd small changes: fix datetime.d * format with comma formatspec to add separator into numbers * Link to ConvException docs in std.bigint ctor doc * apply changes for netbsd: stdio - set name to not null value for reopen * issue 15645 - Prevent unsafe usage of Tuple.slice * std.getopt: Add unit tests for delegates as callbacks. * std.getopt unit tests: No trailing whitespace. * std.getopt unit tests: No trailing whitespace. * Add missing sig constraints to std.numeric.gcd. First of all, it makes no sense to take arbitrary types (e.g., gcd("a","abc") should not even match the template). Second of all, the current implementation only works with built-in types, so the function should restrict itself to only accepting built-in types so that it is overloadable for other types (e.g., BigInt). * Implement generic gcd for non-builtin types. * Documentation. * std.getopt: Make private functions safe. * Annotate unittests per Phobos coding style. * std.getopt: make setConfig safe, pure, nothrow, nogc * SharedFreeList.reallocate must be shared and take a ref void[] * Add ISharedAllocator and have the _processAllocator use it * allow precise scanning of DATA/TLS segment: buffers for preallocated class instances should be typed void[], not ubyte[] and must be properly aligned * workaround ld rejecting explicit alignment * Move std/datetime.d into std/datetime/package.d. * Create the new (empty) files in std/datetime for the split. * Move several items from std/datetime/package.d to std/datetime/common.d. * Move Interval to std.datetime.interval. * Move PosInfInterval to std.datetime.interval. * Move NegInfInterval to std.datetime.interval. * Move interval functions to std.datetime.interval. * Move IntervalRange to std.datetime.interval. * Move PosInfIntervalRange to std.datetime.interval. * Move NegInfInterval to std.datetime.interval. * Move TimeZone to std.datetime.timezone. * Move LocalTime to std.datetime.timezone. * Move UTC to std.datetime.timezone. * Move SimpleTimeZone to std.datetime.timezone. * Move PosixTimeZone to std.datetime.timezone. * Move WindowsTimeZone to std.datetime.timezone. * Move timezone-related free functions to std.datetime.timezone. * Move TimeOfDay to std.datetime.timeofday. * Move Date to std.datetime.date. * Move DateTime to std.datetime.datetime. * [BOOKTABLES]: Add BOOKTABLE to std.encoding * std.range: Add missing methods to the booktable, disable the quickindex and improve the docs * [BOOKTABLES]: Add BOOKTABLE to std.uni * issue# 17372: icmp chokes on a lambda that can throw. This removes some inappropriate attributes from a fullCasedCmp. It's templated, and their applicability depends on the template argument, so they should be inferred. And the unit tests for icmp which test the attributes (including @safe) pass without the attributes on fullCasedCmp, so I have no idea why any of them (particularly @trusted) were ever there. * Split style target for CircleCi * Move SysTime to std.datetime.systime. * Move Clock to std.datetime.systime. * Move stray SysTime-related functions to std.datetime.systime. * Remove some unused private functions. * Fix links in std.datetime. * Fixes to make circleci happy. * Specify the exact folder location for the style Makefile target of the D source files * Add changelog entry for splitting std.datetime. * Move TimeOfDay to std.datetime.date. * Move DateTime to std.datetime.date. * Move stray functionality in std.datetime.common to std.datetime.date. * Remove uses of common, datetime, and timeofday. * Remove std.datetime.common,datetime,timeofday. * Fix links in std.datetime. * Update changelog entry for splitting std.datetime. * Fix Ddoc links in std.datetime * Add the MonoTime equivalents of std.datetime.StopWatch/benchmark. std.datetime.package has StopWatch, benchmark, comparingBenchmark, and measureTime, all of which use TickDuration (which would be deprecated, but it can't be deprecated as long as those functions in std.datetime are deprecated). This commit introduces std.datetime.stopwatch to replace those functions in std.datetime. In order to avoid symbol conflicts, std.datetime.stopwatch will not be publicly import in std.datetime.package until the old symbols have been removed. std.datetime.experimental.stopwatch contains StopWatch and benchmark which have essentially the same APIs as the ones in std.datetime.package, but they use MonoTime and Duration. comparingBenchmark has not been ported to MonoTime and Duration, because it is simply a wrapper around benchmark. measureTime has not been ported to MonoTime and Duration, because it is equivalent to using StopWatch with a scope(exit) statement. The old functionality will be deprecated the major release after the new symbols have been introduced. * Update the changelog entry for splitting std.datetime. Now, it includes information on std.datetime.stopwatch. * Update changelog entry to use MREF instead of LINK2. * Warn about impending deprecation of functions in std.datetime.package. * Logger sharedLog comment on thread-safety fix Issue 16232 - std.experimental.logger.core.sharedLog isn't thread-safe * Move deprecations along. * remove .deps file generation - for development, people can use the much faster std/algorithm.test targets - for batch unittest recompiling isn't of much interest as all the testers just start from a clean slate anyhow - furthermore the .deps generation is broken anyhow, see Issue 7016 * posix.mak: Discard DDoc output of DDoc warning test Send documentation output to /dev/null when checking for warnings. * Fix std.datetime autotester failure for FreeBSD 10.3/11. The tests fail depending on your timezone due to a known FreeBSD bug, so we need to disable them until the bug in FreeBSD gets fixed. * Removed level of indirection from internal std.xml.ElementParser code * Allow parallel execution of public unittests * Check for an existent tools repository on more files * Make behavior of environment.opIndexAssign on Posix to match the behavior on Windows. Setting null as environment value leads to removing this variable according to SetEnvironmentVariable documentation while Posix leaves unspecified what setenev should do if value is null. * Add @safe to two common functions in std.xml * Debugger target for posix It is desirable to be able to attach a debugger to the execution of tests of phobos. This patch add this feature. By calling make -f posix.mak DEBUGGER=gdb std/XXXXX.debug phobos and the module XXXXX will be compiled with debug symbols and after its compilation the debugger gdb starts the compiled executable. DEBUGGER=gdb can be omitted as it is the default debugger. * Fixed some typos in std.string docs * [Static if] replace overload constraints with static if (mutation.d) * Deprecate obsolete pattern matching functions in std.string * Fix some bad/slangy English in std.getopt. I don't think that it makes sense to use slang like "dunno" in a standard library. * Fix issue 17394 - mkdirRecurse isn't @safe * Issue 16053: Fix it so that SysTime's from*String supports more than 7 digits. ISO 8601 says that it's up to the application to decide how many digits to put in the fractional seconds if they're present. SysTime.to*String puts up to 7 (stripping trailing zeroes), because that's hecto-nanosecond precision, and SysTime holds the time in hecto-nanoseconds. Currently, from*String only accepts up to 7 digits in the fractional seconds, which _does_ follow the spec in that (per the spec) the number of digits is up to the applications. However, while we never emit more than 7 digits, other applications do, so only accepting 7 digits makes us incompatible with them, whereas accepting them would make us more compatible with other programs, and it would actually be more efficient, since we'd have fewer checks in the code. So, these changes make is so that SysTime.from*String accepts more than 7 digits in the fractional seconds, but the additional digits are truncated (since SysTime doesn't support more than 7 digits of precision). * Add changelog entry for deprecation of pattern functions * Fix issue 16246 - cannot call iota with 3 [u]bytes or 3 [u]shorts * Fix Issue 16326 - filter is not lazy enough & has weird save behavior * Improve behavior on overflow * std.datetime.stopwatch: Fix random test failure on Win32 * Fix Issue 15720 - iota(long.max, long.min, step) does not work properly * Small clean-up for PR dlang#4027 * Rename slides to slide * Rename slideWithLessElements to withFewerElements * Fixed spelling mistakes in std.range * std.datetime.stopwatch: Fix another random test failure on Win32 * Defer addition of template until we need it for ranges. * Add asOriginalType function for enums

wilzbach · 2017-12-18T15:44:57Z

@ all: I just revived std.range.slide from the dead -> #5943 (and with a bit of luck it might be part of 2.079).

wilzbach force-pushed the pairwise branch from ace80bd to 798fd48 Compare February 29, 2016 23:28

wilzbach changed the title ~~add pairwise range iterator~~ add sliding window range iterator Feb 29, 2016

burner reviewed Mar 2, 2016
View reviewed changes

wilzbach force-pushed the pairwise branch from 798fd48 to e9caa58 Compare March 2, 2016 18:09

JackStouffer reviewed Mar 6, 2016
View reviewed changes

wilzbach force-pushed the pairwise branch from e9caa58 to d1b8638 Compare March 6, 2016 20:03

wilzbach force-pushed the pairwise branch from d1b8638 to 069408c Compare March 15, 2016 10:41

wilzbach changed the title ~~add sliding window range iterator~~ slides: a sliding window range iterator Mar 15, 2016

JakobOvrum reviewed Mar 18, 2016
View reviewed changes

wilzbach force-pushed the pairwise branch 2 times, most recently from ff9e238 to c2b1e68 Compare March 23, 2016 22:59

9il added Review:Needs Decision Review:Needs Work labels Mar 25, 2016

wilzbach force-pushed the pairwise branch from 28e44fe to 12fd0e2 Compare February 16, 2017 23:41

wilzbach removed the Review:Needs Work label Feb 16, 2017

Fix issue 6621 - add sliding window iterator

db2bf8f

wilzbach force-pushed the pairwise branch from 12fd0e2 to db2bf8f Compare February 17, 2017 00:17

JackStouffer suggested changes Feb 17, 2017

View reviewed changes

wilzbach added this to the 2.074.0 milestone Feb 18, 2017

wilzbach added the Review:Needs Work label Feb 19, 2017

wilzbach removed this from the 2.074.0 milestone Feb 19, 2017

JackStouffer added this to the 2.074.0 milestone Feb 22, 2017

wilzbach modified the milestones: 2.075.0, 2.074.0 Mar 4, 2017

Merge branch 'master' into pairwise

f92adce

JackStouffer approved these changes May 15, 2017

View reviewed changes

JackStouffer added Merge:auto-merge and removed Review:Needs Work labels May 15, 2017

dlang-bot merged commit e872190 into dlang:master May 15, 2017

JackStouffer added a commit to JackStouffer/phobos that referenced this pull request May 15, 2017

Small clean-up for PR dlang#4027

575ed52

andralex pushed a commit to andralex/phobos that referenced this pull request May 16, 2017

Merge pull request dlang#4027 from wilzbach/pairwise

be63fc7

slides: a sliding window range iterator merged-on-behalf-of: unknown

CyberShadow mentioned this pull request May 16, 2017

Fix Issue 15720 - iota(long.max, long.min, step) does not work properly #5396

Closed

JackStouffer added a commit to JackStouffer/phobos that referenced this pull request May 16, 2017

Small clean-up for PR dlang#4027

cc44bc4

dlang-bot added a commit that referenced this pull request May 16, 2017

Merge pull request #5395 from JackStouffer/slides-clean-up

4dab22e

Clean-up for PR #4027 merged-on-behalf-of: Andrei Alexandrescu <andralex@users.noreply.github.com>

wilzbach deleted the pairwise branch July 15, 2017 23:18

wilzbach mentioned this pull request Jul 15, 2017

Delay release of std.range.slides until 2.076 #5614

Merged

wilzbach mentioned this pull request Dec 18, 2017

Revive std.range.slide from the dead #5943

Merged

Uh oh!

Conversation

wilzbach commented Feb 27, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Why is this useful?

Current alternatives

Trivia

Uh oh!

yebblies commented Feb 28, 2016

Uh oh!

wilzbach commented Feb 29, 2016

Scala: sliding

Ruby: each_cons

Clojure: partition

Uh oh!

quickfur commented Feb 29, 2016

Uh oh!

wilzbach commented Feb 29, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

wilzbach commented Mar 4, 2016

Uh oh!

wilzbach commented Mar 4, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

wilzbach commented Mar 8, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rcorre commented Mar 20, 2016

Uh oh!

wilzbach commented Mar 23, 2016

Uh oh!

JakobOvrum commented Mar 23, 2016

Uh oh!

wilzbach commented Mar 25, 2016

Uh oh!

9il commented Mar 25, 2016

Uh oh!

schveiguy commented Mar 25, 2016

Uh oh!

wilzbach commented Feb 16, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

JackStouffer left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

JackStouffer commented May 5, 2017

Uh oh!

JackStouffer commented May 15, 2017

Uh oh!

wilzbach commented Dec 18, 2017

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

wilzbach commented Feb 27, 2016 •

edited

Loading

wilzbach commented Feb 16, 2017 •

edited

Loading