Skip to content

Latest commit

 

History

History
957 lines (758 loc) · 38.7 KB

changelog_1_6_0.md

File metadata and controls

957 lines (758 loc) · 38.7 KB

v1.6.0 - 2021-10-14

Major new features

With so many new features, pinpointing the most salient ones is a subjective exercise, but here are a select few:

iterable[T]

The iterable[T] type class was added to match called iterators, which solves a number of long-standing issues related to iterators. Example:

iterator iota(n: int): int =
  for i in 0..<n: yield i

# previously, you'd need `untyped`, which caused other problems such as lack
# of type inference, overloading issues, and MCS.
template sumOld(a: untyped): untyped = # no type inference possible
  var result: typeof(block:(for ai in a: ai))
  for ai in a: result += ai
  result

assert sumOld(iota(3)) == 0 + 1 + 2

# now, you can write:
template sum[T](a: iterable[T]): T =
  # `template sum(a: iterable): auto =` would also be possible
  var result: T
  for ai in a: result += ai
  result

assert sum(iota(3)) == 0 + 1 + 2 # or `iota(3).sum`

In particular iterable arguments can now be used with the method call syntax. For example:

import std/[sequtils, os]
echo walkFiles("*").toSeq # now works

See PR #17196 for additional details.

Strict effects

The effect system was refined and there is a new .effectsOf annotation that does explicitly what was previously done implicitly. See the manual for more details. To write code that is portable with older Nim versions, use this idiom:

when defined(nimHasEffectsOf):
  {.experimental: "strictEffects".}
else:
  {.pragma: effectsOf.}

proc mysort(s: seq; cmp: proc(a, b: T): int) {.effectsOf: cmp.}

To enable the new effect system, compile with --experimental:strictEffects. See also #18777 and RFC #408.

Private imports and private field access

A new import syntax import foo {.all.} now allows importing all symbols (public or private) from foo. This can be useful for testing purposes or for more flexibility in project organization.

Example:

from system {.all.} as system2 import nil
echo system2.ThisIsSystem # ThisIsSystem is private in `system`
import os {.all.} # weirdTarget is private in `os`
echo weirdTarget # or `os.weirdTarget`

Added a new module std/importutils, and an API privateAccess, which allows access to private fields for an object type in the current scope.

Example:

import times
from std/importutils import privateAccess
block:
  let t = now()
  # echo t.monthdayZero # Error: undeclared field: 'monthdayZero' for type times.DateTime
  privateAccess(typeof(t)) # enables private access in this scope
  echo t.monthdayZero # ok

See PR #17706 for additional details.

nim --eval:cmd

Added nim --eval:cmd to evaluate a command directly, e.g.: nim --eval:"echo 1". It defaults to e (nimscript) but can also work with other commands, e.g.:

find . | nim r --eval:'import strutils; for a in stdin.lines: echo a.toUpper'
# use as a calculator:
nim --eval:'echo 3.1 / (1.2+7)'
# explore a module's APIs, including private symbols:
nim --eval:'import os {.all.}; echo weirdTarget'
# use a custom backend:
nim r -b:js --eval:"import std/jsbigints; echo 2'big ** 64'big"

See PR #15687 for more details.

Round-trip float to string

system.addFloat and system.$ now can produce string representations of floating point numbers that are minimal in size and possess round-trip and correct rounding guarantees (via the Dragonbox algorithm). This currently has to be enabled via -d:nimPreviewFloatRoundtrip. It is expected that this behavior becomes the new default in upcoming versions, as with other nimPreviewX define flags.

Example:

from math import round
let a = round(9.779999999999999, 2)
assert a == 9.78
echo a # with `-d:nimPreviewFloatRoundtrip`: 9.78, like in python3 (instead of  9.779999999999999)

New std/jsbigints module

Provides arbitrary precision integers for the JS target. See PR #16409. Example:

import std/jsbigints
assert 2'big ** 65'big == 36893488147419103232'big
echo 0xdeadbeef'big shl 4'big # 59774856944n

New std/sysrand module

Cryptographically secure pseudorandom number generator, allows generating random numbers from a secure source provided by the operating system. Example:

import std/sysrand
assert urandom(1234) != urandom(1234) # unlikely to fail in practice

See PR #16459.

New module: std/tempfiles

Allows creating temporary files and directories, see PR #17361 and followups.

import std/tempfiles
let tmpPath = genTempPath("prefix", "suffix.log", "/tmp/")
# tmpPath looks like: /tmp/prefixpmW1P2KLsuffix.log

let dir = createTempDir("tmpprefix_", "_end")
# created dir looks like: getTempDir() / "tmpprefix_YEl9VuVj_end"

let (cfile, path) = createTempFile("tmpprefix_", "_end.tmp")
# path looks like: getTempDir() / "tmpprefix_FDCIRZA0_end.tmp"
cfile.write "foo"
cfile.setFilePos 0
assert readAll(cfile) == "foo"
close cfile
assert readFile(path) == "foo"

User-defined literals

Custom numeric literals (e.g. -128'bignum) are now supported. Additionally, the unary minus in -1 is now part of the integer literal, i.e. it is now parsed as a single token. This implies that edge cases like -128'i8 finally work correctly. Example:

func `'big`*(num: cstring): JsBigInt {.importjs: "BigInt(#)".}
assert 0xffffffffffffffff'big == (1'big shl 64'big) - 1'big

Dot-like operators

With -d:nimPreviewDotLikeOps, dot-like operators (operators starting with ., but not with ..) now have the same precedence as ., so that a.?b.c is now parsed as (a.?b).c instead of a.?(b.c). A warning is generated when a dot-like operator is used without -d:nimPreviewDotLikeOps.

An important use case is to enable dynamic fields without affecting the built-in . operator, e.g. for std/jsffi, std/json, pkg/nimpy. Example:

import std/json
template `.?`(a: JsonNode, b: untyped{ident}): JsonNode =
  a[astToStr(b)]
let j = %*{"a1": {"a2": 10}}
assert j.?a1.?a2.getInt == 10

Block arguments now support optional parameters

This solves a major pain point for routines accepting block parameters, see PR #18631 for details:

template fn(a = 1, b = 2, body) = discard
fn(1, 2): # already works
  bar
fn(a = 1): # now works
  bar

Likewise with multiple block arguments via do:

template fn(a = 1, b = 2, body1, body2) = discard
fn(a = 1): # now works
  bar1
do:
  bar2

Other new features

New and deprecated modules

The following modules were added (they are discussed in the rest of the text):

  • std/enumutils

  • std/genasts

  • std/importutils

  • std/jsbigints

  • std/jsfetch

  • std/jsformdata

  • std/jsheaders

  • std/packedsets

  • std/setutils

  • std/socketstreams

  • std/strbasics

  • std/sysrand

  • std/tasks

  • std/tempfiles

  • std/vmutils

  • Deprecated std/mersenne.

  • Removed deprecated std/iup module from stdlib; it has already moved to nimble.

New std/jsfetch module

Provides a wrapper for JS Fetch API. Example:

# requires -d:nimExperimentalAsyncjsThen
import std/[jsfetch, asyncjs, jsconsole, jsffi, sugar]
proc fn {.async.} =
  await fetch("https://api.github.com/users/torvalds".cstring)
    .then((response: Response) => response.json())
    .then((json: JsObject) => console.log(json))
    .catch((err: Error) => console.log("Request Failed", err))
discard fn()

New std/tasks module

Provides basic primitives for creating parallel programs. Example:

import std/tasks
var num = 0
proc hello(a: int) = num+=a

let b = toTask hello(13) # arguments must be isolated, see `std/isolation`
b.invoke()
assert num == 13
b.invoke() # can be called again
assert num == 26

New module: std/genasts

Provides an API genAst that avoids the problems inherent with quote do and can be used as a replacement. Example showing how this could be used for writing a simplified version of unittest.check:

import std/[genasts, macros, strutils]
macro check2(cond: bool): untyped =
  assert cond.kind == nnkInfix, "$# not implemented" % $cond.kind
  result = genAst(cond, s = repr(cond), lhs = cond[1], rhs = cond[2]):
    # each local symbol we access must be explicitly captured
    if not cond:
      doAssert false, "'$#'' failed: lhs: '$#', rhs: '$#'" % [s, $lhs, $rhs]
let a = 3
check2 a*2 == a+3
if false: check2 a*2 < a+1 # would error with: 'a * 2 < a + 1'' failed: lhs: '6', rhs: '4'

See PR #17426 for details.

New module: std/setutils

  • Added setutils.toSet that can take any iterable and convert it to a built-in set, if the iterable yields a built-in settable type.
  • Added setutils.fullSet which returns a full built-in set for a valid type.
  • Added setutils.complement which returns the complement of a built-in set.
  • Added setutils.[]=.

New module: std/enumutils

  • Added genEnumCaseStmt macro that generates case statement to parse string to enum.
  • Added items for enums with holes.
  • Added symbolName to return the enum symbol name ignoring the human-readable name.
  • Added symbolRank to return the index in which an enum member is listed in an enum.

system

  • Added system.prepareMutation for better support of low level moveMem, copyMem operations for gc:orc's copy-on-write string implementation.
  • system.addEscapedChar now renders \r as \r instead of \c, to be compatible with most other languages.
  • Added cmpMem to system.
  • doAssertRaises now correctly handles foreign exceptions.
  • addInt now supports unsigned integers.

Compatibility notes:

  • system.delete had surprising behavior when the index passed to it was out of bounds (it would delete the last entry then). Compile with -d:nimStrictDelete so that an index error is produced instead. Be aware however that your code might depend on this quirky behavior so a review process is required on your part before you can use -d:nimStrictDelete. To make this review easier, use the -d:nimAuditDelete switch, which pretends that system.delete is deprecated so that it is easier to see where it was used in your code. -d:nimStrictDelete will become the default in upcoming versions.
  • cuchar is now deprecated as it aliased char where arguably it should have aliased uint8. Please use char or uint8 instead.
  • repr now doesn't insert trailing newlines; the previous behavior was very inconsistent, see #16034. Use -d:nimLegacyReprWithNewline for the previous behavior. repr now also renders ASTs correctly for user defined literals, setters, do, etc.
  • Deprecated any. See RFC #281.
  • The unary slice ..b was deprecated, use 0..b instead.

std/math

  • Added almostEqual for comparing two float values using a machine epsilon.
  • Added clamp which allows using a Slice to clamp to a value.
  • Added ceilDiv for integer division that rounds up.
  • Added isNaN.
  • Added copySign.
  • Added euclDiv and euclMod.
  • Added signbit.
  • Added frexp overload procs. Deprecated c_frexp, use frexp instead.

Compatibility notes:

  • math.round now rounds "away from zero" in the JS backend, which is consistent with other backends. See #9125. Use -d:nimLegacyJsRound for the previous behavior.

Random number generators: std/random, std/sysrand, std/oids

  • Added std/sysrand module (see details above).
  • Added randState template that exposes the default random number generator. Useful for library authors.
  • Added initRand() overload with no argument which uses the current time as a seed.
  • initRand(seed) now allows seed == 0.
  • Fixed overflow bugs.
  • Fix initRand to avoid random number sequences overlapping, refs #18744.
  • std/oids now uses std/random.

Compatibility notes:

  • Deprecated std/mersenne.
  • random.initRand(seed) now produces non-skewed values for the first call to rand() after initialization with a small (< 30000) seed. Use -d:nimLegacyRandomInitRand to restore previous behavior for a transition time, see PR #17467.

std/json, std/jsonutils

  • With -d:nimPreviewJsonutilsHoleyEnum, jsonutils now can serialize/deserialize holey enums as regular enums (via ord) instead of as strings. It is expected that this behavior becomes the new default in upcoming versions. toJson now serializes JsonNode as is via reference (without a deep copy) instead of treating JsonNode as a regular ref object, this can be customized via jsonNodeMode.
  • std/json and std/jsonutils now serialize NaN, Inf, -Inf as strings, so that %[NaN, -Inf] is the string ["nan","-inf"] instead of [nan,-inf] which was invalid JSON.
  • std/json can now handle integer literals and floating point literals of arbitrary length and precision. Numbers that do not fit the underlying BiggestInt or BiggestFloat fields are kept as string literals and one can use external BigNum libraries to handle these. The parseFloat family of functions also has now optional rawIntegers and rawFloats parameters that can be used to enforce that all integer or float literals remain in the "raw" string form so that client code can easily treat small and large numbers uniformly.
  • Added BackwardsIndex overload for JsonNode.
  • json.%,json.to, jsonutils.fromJson,jsonutils.toJson now work with uint|uint64 instead of raising (as in 1.4) or giving wrong results (as in 1.2).
  • std/jsonutils now handles cstring (including as Table key), and set.
  • Added jsonutils.jsonTo overload with opt = Joptions() param.
  • jsonutils.toJson now supports customization via ToJsonOptions.
  • std/json, std/jsonutils now support round-trip serialization when -d:nimPreviewFloatRoundtrip is used.

std/typetraits, std/compilesettings

  • distinctBase now is identity instead of error for non distinct types.
  • distinctBase now allows controlling whether to be recursive or not.
  • Added enumLen to return the number of elements in an enum.
  • Added HoleyEnum for enums with holes, OrdinalEnum for enums without holes.
  • Added hasClosure.
  • Added pointerBase to return T for ref T | ptr T.
  • Added compilesettings.SingleValueSetting.libPath.

networking: std/net, std/asyncnet, std/htmlgen, std/httpclient, std/asyncdispatch, std/asynchttpserver, std/httpcore

  • Fixed buffer overflow bugs in std/net.
  • Exported sslHandle from std/net and std/asyncnet.
  • Added hasDataBuffered to std/asyncnet.
  • Various functions in std/httpclient now accept url of type Uri. Moreover, the request function's httpMethod argument of type string was deprecated in favor of HttpMethod enum type; see #15919.
  • Added asyncdispatch.activeDescriptors that returns the number of currently active async event handles/file descriptors.
  • Added getPort to std/asynchttpserver to resolve OS-assigned Port(0); this is usually recommended instead of hardcoding ports which can lead to "Address already in use" errors.
  • Fixed premature garbage collection in std/asyncdispatch, when a stacktrace override is in place.
  • Added httpcore.is1xx and missing HTTP codes.
  • Added htmlgen.portal for making "SPA style" pages using HTML only.

Compatibility notes:

  • On Windows, the SSL library now checks for valid certificates. For this purpose it uses the cacert.pem file, which was extracted from https://curl.se/ca/cacert.pem. Besides the OpenSSL DLLs (e.g. libssl-1_1-x64.dll, libcrypto-1_1-x64.dll) you now also need to ship cacert.pem with your .exe file.

std/hashes

  • hashes.hash can now support object and ref (can be overloaded in user code), if -d:nimPreviewHashRef is used. It is expected that this behavior becomes the new default in upcoming versions.
  • hashes.hash(proc|ptr|ref|pointer) now calls hash(int) and honors -d:nimIntHash1. hashes.hash(closure) has also been improved.

OS: std/os, std/io, std/socketstream, std/linenoise, std/tempfiles

  • os.FileInfo (returned by getFileInfo) now contains blockSize, determining preferred I/O block size for this file object.
  • Added os.getCacheDir() to return platform specific cache directory.
  • Improved os.getTempDir(), see PR #16914.
  • Added os.isAdmin to tell whether the caller's process is a member of the Administrators local group (on Windows) or a root (on POSIX).
  • Added optional options argument to copyFile, copyFileToDir, and copyFileWithPermissions. By default, on non-Windows OSes, symlinks are followed (copy files symlinks point to); on Windows, options argument is ignored and symlinks are skipped.
  • On non-Windows OSes, copyDir and copyDirWithPermissions copy symlinks as symlinks (instead of skipping them as it was before); on Windows symlinks are skipped.
  • On non-Windows OSes, moveFile and moveDir move symlinks as symlinks (instead of skipping them sometimes as it was before).
  • Added optional followSymlinks argument to setFilePermissions.
  • Added a simpler to use io.readChars overload.
  • Added socketstream module that wraps sockets in the stream interface.
  • Added experimental linenoise.readLineStatus to get line and status (e.g. ctrl-D or ctrl-C).

Environment variable handling

  • Empty environment variable values are now supported across OS's and backends.
  • Environment variable APIs now work in multithreaded scenarios, by delegating to direct OS calls instead of trying to keep track of the environment.
  • putEnv, delEnv now work at compile time.
  • NodeJS backend now supports osenv: getEnv, putEnv, envPairs, delEnv, existsEnv.

Compatibility notes:

  • std/os: putEnv now raises if the first argument contains a =.

POSIX

  • On POSIX systems, the default signal handlers used for Nim programs (it's used for printing the stacktrace on fatal signals) will now re-raise the signal for the OS default handlers to handle. This lets the OS perform its default actions, which might include core dumping (on select signals) and notifying the parent process about the cause of termination.
  • On POSIX systems, we now ignore SIGPIPE signals, use -d:nimLegacySigpipeHandler for previous behavior.
  • Added posix_utils.osReleaseFile to get system identification from os-release file on Linux and the BSDs. (link)
  • Removed undefined behavior for posix.open.

std/prelude

  • std/strformat is now part of include std/prelude.
  • Added std/sequtils import to std/prelude.
  • std/prelude now works with the JS target.
  • std/prelude can now be used via include std/prelude, but include prelude still works.

String manipulation: std/strformat, std/strbasics

  • Added support for parenthesized expressions.
  • Added support for const strings instead of just string literals.
  • Added std/strbasics for high-performance string operations.
  • Added strip, setSlice, add(a: var string, b: openArray[char]).

std/wrapnils

  • std/wrapnils doesn't use experimental:dotOperators anymore, avoiding issues like bug #13063 (which affected error messages) for modules importing std/wrapnils.
  • Added ??. macro which returns an Option.
  • std/wrapnils can now be used to protect against FieldDefect errors in case objects, generates optimal code (no overhead compared to manual if-else branches), and preserves lvalue semantics which allows modifying an expression.

Containers: std/algorithm, std/lists, std/sequtils, std/options, std/packedsets

  • Removed the optional longestMatch parameter of the critbits._WithPrefix iterators (it never worked reliably).
  • Added algorithm.merge.
  • In std/lists: renamed append to add and retained append as an alias; added prepend and prependMoved analogously to add and addMoved; added remove for SinglyLinkedLists.
  • Added new operations for singly- and doubly linked lists: lists.toSinglyLinkedList and lists.toDoublyLinkedList convert from openArrays; lists.copy implements shallow copying; lists.add concatenates two lists - an O(1) variation that consumes its argument, addMoved, is also supplied. See PRs #16362, #16536.
  • New module: std/packedsets. Generalizes std/intsets, see PR #15564.

Compatibility notes:

  • Deprecated sequtils.delete and added an overload taking a Slice that raises a defect if the slice is out of bounds, likewise with strutils.delete.
  • Deprecated proc reversed*[T](a: openArray[T], first: Natural, last: int): seq[T] in std/algorithm.
  • std/options changed $some(3) to "some(3)" instead of "Some(3)" and $none(int) to "none(int)" instead of "None[int]".

std/times

  • Added ZZZ and ZZZZ patterns to times.nim DateTime parsing, to match time zone offsets without colons, e.g. UTC+7 -> +0700.
  • Added dateTime and deprecated initDateTime.

std/macros and AST

  • New module std/genasts, see description above.
  • The required name of case statement macros for the experimental caseStmtMacros feature has changed from match to `case`.
  • Tuple expressions are now parsed consistently as nnkTupleConstr node. Will affect macros expecting nodes to be of nnkPar.
  • In std/macros, treeRepr,lispRepr,astGenRepr now represent SymChoice nodes in a collapsed way. Use -d:nimLegacyMacrosCollapseSymChoice to get the previous behavior.
  • Made custom op in macros.quote work for all statements.

std/sugar

  • Added sugar.dumpToString which improves on sugar.dump.
  • Added an overload for the collect macro that infers the container type based on the syntax of the last expression. Works with std seqs, tables and sets.

Compatibility notes:

  • Removed support for named procs in sugar.=>.

Parsing: std/parsecfg, std/strscans, std/uri

  • Added sections iterator in parsecfg.
  • strscans.scanf now supports parsing single characters.
  • Added strscans.scanTuple which uses strscans.scanf internally, returning a tuple which can be unpacked for easier usage of scanf.
  • Added decodeQuery to std/uri.
  • parseopt.initOptParser has been made available and parseopt has been added back to std/prelude for all backends. Previously initOptParser was unavailable if the std/os module did not have paramCount or paramStr, but the use of these in initOptParser were conditionally to the runtime arguments passed to it, so initOptParser has been changed to raise ValueError when the real command line is not available. parseopt was previously excluded from std/prelude for JS, as it could not be imported.

Compatibility notes:

  • Changed the behavior of uri.decodeQuery when there are unencoded = characters in the decoded values. Prior versions would raise an error. This is no longer the case to comply with the HTML spec and other languages' implementations. Old behavior can be obtained with -d:nimLegacyParseQueryStrict. cgi.decodeData which uses the same underlying code is also updated the same way.

JS stdlib changes

  • Added std/jsbigints module, which provides arbitrary precision integers for the JS target.
  • Added setCurrentException for the JS backend.
  • writeStackTrace is available in the JS backend now.
  • Added then, catch to std/asyncjs for promise pipelining, for now hidden behind -d:nimExperimentalAsyncjsThen.
  • Added std/jsfetch module Fetch wrapper for the JS target.
  • Added std/jsheaders module Headers wrapper for the JS target.
  • Added std/jsformdata module FormData wrapper for the JS target.
  • Added jscore.debugger to call any available debugging functionality, such as breakpoints.
  • Added jsconsole.dir, jsconsole.dirxml, jsconsole.timeStamp.
  • Added dollar $ and len for jsre.RegExp.
  • Added jsconsole.jsAssert for the JS target.
  • Added ** to std/jsffi.
  • Added copyWithin for seq and array for JS targets.
  • In std/dom, Interval is now a ref object, same as Timeout. Definitions of setTimeout, clearTimeout, setInterval, clearInterval were updated.
  • Added dom.scrollIntoView proc with options.
  • Added dom.setInterval, dom.clearInterval overloads.
  • Merged std/dom_extensions into the std/dom module, as it was a module with a single line, see RFC #413.
  • $ now gives more correct results on the JS backend.

JS compiler changes

  • cstring doesn't support the []= operator anymore in the JS backend.
  • Array literals now use JS typed arrays when the corresponding JS typed array exists, for example [byte(1), 2, 3] generates new Uint8Array([1, 2, 3]).

VM and nimscript backend

  • VM now supports addr(mystring[ind]) (index + index assignment).
  • nimscript now handles except Exception as e.
  • nil dereference is not allowed at compile time. cast[ptr int](nil)[] is rejected at compile time.
  • static[T] now works better, refs #17590, #15853.
  • distinct T conversions now work in VM.
  • items(cstring) now works in VM.
  • Fix addr, len, high in VM (#16002, #16610).
  • std/cstrutils now works in VM.

OS/platform-specific notes

  • Support for Apple silicon/M1.
  • Support for 32-bit RISC-V, refs #16231.
  • Support for armv8l, refs #18901.
  • Support for CROSSOS, refs #18889.
  • The allocator for Nintendo Switch, which was nonfunctional because of breaking changes in libnx, was removed, in favor of the new -d:nimAllocPagesViaMalloc option.
  • Allow reading parameters when compiling for Nintendo Switch.
  • --nimcache now correctly works in a cross-compilation setting.
  • Cross compilation targeting Windows was improved.
  • This now works from macOS/Linux: nim r -d:mingw main

Performance / memory optimizations

  • The comment field in PNode AST was moved to a side channel, reducing overall memory usage during compilation by a factor 1.25x
  • std/jsonutils deserialization is now up to 20x faster.
  • os.copyFile is now 2.5x faster on macOS, by using copyfile from copyfile.h; use -d:nimLegacyCopyFile for macOS < 10.5.
  • Float to string conversion is now 10x faster thanks to the Dragonbox algorithm, with -d:nimPreviewFloatRoundtrip.
  • newSeqWith is 3x faster.
  • CI now supports batching (making Windows CI 2.3X faster).
  • Sets now uses the optimized countSetBits proc, see PR #17334.

Debugging

  • You can now enable/disable VM tracing in user code via vmutils.vmTrace.
  • koch tools now builds bin/nim_dbg which allows easy access to a debug version of Nim without recompiling.
  • Added new module compiler/debugutils to help with debugging Nim compiler.
  • Renamed -d:nimCompilerStackraceHints to -d:nimCompilerStacktraceHints and used it in more contexts; this flag which works in tandem with --stackTraceMsgs to show user code context in compiler stacktraces.

Type system

  • typeof(voidStmt) now works and returns void.
  • enum values can now be overloaded. This needs to be enabled via {.experimental: "overloadableEnums".}. We hope that this feature allows for the development of more fluent (less ugly) APIs. See RFC #373 for more details.
  • A type conversion from one enum type to another now produces an [EnumConv] warning. You should use ord (or cast, but the compiler won't help, if you misuse it) instead.
    type A = enum a1, a2
    type B = enum b1, b2
    echo a1.B # produces a warning
    echo a1.ord.B # produces no warning
  • A dangerous implicit conversion to cstring now triggers a [CStringConv] warning. This warning will become an error in future versions! Use an explicit conversion like cstring(x) in order to silence the warning.
  • There is a new warning for any type conversion to enum that can be enabled via .warning[AnyEnumConv]:on or --warning:AnyEnumConv:on.
  • Reusing a type name in a different scope now works, refs #17710.
  • Fixed implicit and explicit generics in procedures, refs #18808.

New-style concepts

Example:

type
  Comparable = concept # no T, an atom
    proc cmp(a, b: Self): int

The new design does not rely on system.compiles and may compile faster. See PR #15251 and RFC #168 for details.

Lexical / syntactic

  • Nim now supports a small subset of Unicode operators as operator symbols. The supported symbols are: "∙ ∘ × ★ ⊗ ⊘ ⊙ ⊛ ⊠ ⊡ ∩ ∧ ⊓ ± ⊕ ⊖ ⊞ ⊟ ∪ ∨ ⊔". To enable this feature, use --experimental:unicodeOperators. Note that due to parser limitations you cannot enable this feature via a pragma {.experimental: "unicodeOperators".} reliably, you need to enable it via the command line or in a configuration file.
  • var a {.foo.} = expr now works inside templates (except when foo is overloaded).

Compiler messages, error messages, hints, warnings

  • Significant improvement to error messages involving effect mismatches, see PRs #18384, #18418.
  • Added --declaredLocs to show symbol declaration location in error messages.
  • Added --spellSuggest to show spelling suggestions on typos.
  • Added --processing:dots|filenames|off which customizes hintProcessing; --processing:filenames shows which include/import modules are being compiled as an import stack.
  • FieldDefect messages now shows discriminant value + lineinfo, in all backends (C, JS, VM)
  • Added --hintAsError with similar semantics as --warningAsError.
  • Added --unitsep:on|off to control whether to add ASCII unit separator \31 before a newline for every generated message (potentially multiline), so tooling can tell when messages start and end.
  • Added --filenames:abs|canonical|legacyRelProj which replaces --listFullPaths:on|off
  • --hint:all:on|off is now supported to select or deselect all hints; it differs from --hints:on|off which acts as a (reversible) gate. Likewise with --warning:all:on|off.
  • The style checking of the compiler now supports a --styleCheck:usages switch. This switch enforces that every symbol is written as it was declared, not enforcing the official Nim style guide. To be enabled, this has to be combined either with --styleCheck:error or --styleCheck:hint.
  • Type mismatch errors now show more context, use -d:nimLegacyTypeMismatch for previous behavior.
  • typedesc[Foo] now renders as such instead of type Foo in compiler messages.
  • runnableExamples now show originating location in stacktraces on failure.
  • SuccessX message now shows more useful information.
  • New DuplicateModuleImport warning; improved UnusedImport and XDeclaredButNotUsed accuracy.

Compatibility notes:

  • --hint:CC now prints to stderr (like all other hints) instead of stdout.

Building and running Nim programs, configuration system

  • JSON build instructions are now generated in $nimcache/outFileBasename.json instead of $nimcache/projectName.json. This allows avoiding recompiling a given project compiled with different options if the output file differs.
  • --usenimcache (implied by nim r main) now generates an output file that includes a hash of some of the compilation options, which allows caching generated binaries:
      nim r main # recompiles
      nim r -d:foo main # recompiles
      nim r main # uses cached binary
      nim r main arg1 arg2 # likewise (runtime arguments are irrelevant)
  • nim r now supports cross compilation from unix to windows when specifying -d:mingw by using Wine, e.g.: nim r --eval:'import os; echo "a" / "b"' prints a\b.
  • nim can compile version 1.4.0 as follows: nim c --lib:lib --stylecheck:off -d:nimVersion140 compiler/nim. -d:nimVersion140 is not needed for bootstrapping, only for building 1.4.0 from devel.
  • nim e now accepts arbitrary file extensions for the nimscript file, although .nims is still the preferred extension in general.
  • The configuration subsystem now allows for -d:release and -d:danger to work as expected. The downside is that these defines now have custom logic that doesn't apply for other defines.

Multithreading

  • TLS: macOS now uses native TLS (--tlsEmulation:off). TLS now works with importcpp non-POD types; such types must use .cppNonPod and --tlsEmulation:offshould be used.
  • Added unsafeIsolate and extract to std/isolation.
  • Added std/tasks, see description above.

Memory management

  • --gc:arc now bootstraps (PR #17342).
  • Lots of improvements to gc:arc, gc:orc, see PR #15697, #16849, #17993.
  • --gc:orc is now 10% faster than previously for common workloads. If you have trouble with its changed behavior, compile with -d:nimOldOrc.
  • The --gc:orc algorithm was refined so that custom container types can participate in the cycle collection process. See the documentation of =trace for more details.
  • On embedded devices malloc can now be used instead of mmap via -d:nimAllocPagesViaMalloc. This is only supported for --gc:orc or --gc:arc.

Compatibility notes:

  • --newruntime and --refchecks are deprecated, use --gc:arc, --gc:orc, or --gc:none as appropriate instead.

Docgen

  • docgen: RST files can now use single backticks instead of double backticks and correctly render in both nim rst2html (as before) as well as common tools rendering RST directly (e.g. GitHub). This is done by adding the default-role:: code directive inside the RST file (which is now handled by nim rst2html).
  • Source+Edit links now appear on top of every docgen'd page when nim doc --git.url:url ... is given.
  • Latex doc generation is revised: output .tex files should be compiled by xelatex (not by pdflatex as before). Now default Latex settings provide support for Unicode and better avoid margin overflows. The minimum required version is TeXLive 2018 (or an equivalent MikTeX version).
  • The RST parser now supports footnotes, citations, admonitions, and short style references with symbols.
  • The RST parser now supports Markdown table syntax. Known limitations:
    • cell alignment is not supported, i.e. alignment annotations in a delimiter row (:---, :--:, ---:) are ignored
    • every table row must start with |, e.g. | cell 1 | cell 2 |.
  • Implemented doc2tex compiler command which converts documentation in .nim files to Latex.
  • docgen now supports syntax highlighting for inline code.
  • docgen now supports same-line doc comments:
    func fn*(a: int): int = 42  ## Doc comment
  • docgen now renders deprecated and other pragmas.
  • runnableExamples now works with templates and nested templates.
  • runnableExamples: "-r:off" now works for examples that should compile but not run.
  • runnableExamples now renders code verbatim, and produces correct code in all cases.
  • docgen now shows correct, canonical import paths in docs.
  • docgen now shows all routines in sidebar, and the proc signature is now shown in sidebar.

Effects and checks

  • Significant improvement to error messages involving effect mismatches
  • There is a new cast section {.cast(uncheckedAssign).}: body that disables some compiler checks regarding case objects. This allows serialization libraries to avoid ugly, non-portable solutions. See RFC #407 for more details.

Compatibility notes:

  • Fixed effect tracking for borrowed procs (see #18882). One consequence is that, under some circumstances, Nim could previously permit a procedure with side effects to be written with func - you may need to change some occurrences of func to proc. To illustrate, Nim versions before 1.6.0 compile the below without error
    proc print(s: string) =
      echo s
    
    type
      MyString = distinct string
    
    proc print(s: MyString) {.borrow.}
    
    func foo(s: MyString) =
      print(s)
    but Nim 1.6.0 produces the error
    Error: 'foo' can have side effects
    
    similar to how we expect that
    func print(s: string) =
      echo s
    produces
    Error: 'print' can have side effects
    

Tools

  • Major improvements to nimgrep, see PR #15612 .
  • fusion is now un-bundled from Nim, ./koch fusion will install it via Nimble at a fixed hash.
  • testament: added nimoutFull: bool spec to compare full output of compiler instead of a subset; many bugfixes to testament.

Misc/cleanups

  • Deprecated TaintedString and --taintmode.
  • Deprecated --nilseqs which is now a noop.
  • Added -d:nimStrictMode in CI in several places to ensure code doesn't have certain hints/warnings.
  • Removed .travis.yml, appveyor.yml.disabled, .github/workflows/ci.yml.disabled.
  • [skip ci] now works in azure and CI pipelines, see detail in PR #17561.