diff --git a/cmd/dumbindent/main.go b/cmd/dumbindent/main.go index e9964dfc..601c6bf9 100644 --- a/cmd/dumbindent/main.go +++ b/cmd/dumbindent/main.go @@ -35,20 +35,20 @@ // To quantify "much faster", on this one C file, this program was 80 times // faster than `clang-format`, even without a column limit: // -// $ wc release/c/wuffs-v0.2.c -// 11858 35980 431885 release/c/wuffs-v0.2.c -// $ time dumbindent < release/c/wuffs-v0.2.c > /dev/null -// real 0m0.008s -// user 0m0.005s -// sys 0m0.005s -// $ time clang-format-9 < release/c/wuffs-v0.2.c > /dev/null -// real 0m0.668s -// user 0m0.618s -// sys 0m0.032s -// $ time clang-format-9 -style='{ColumnLimit: 0}' < release/c/wuffs-v0.2.c > /dev/null -// real 0m0.641s -// user 0m0.585s -// sys 0m0.037s +// $ wc release/c/wuffs-v0.2.c +// 11858 35980 431885 release/c/wuffs-v0.2.c +// $ time dumbindent < release/c/wuffs-v0.2.c > /dev/null +// real 0m0.008s +// user 0m0.005s +// sys 0m0.005s +// $ time clang-format-9 < release/c/wuffs-v0.2.c > /dev/null +// real 0m0.668s +// user 0m0.618s +// sys 0m0.032s +// $ time clang-format-9 -style='{ColumnLimit: 0}' < release/c/wuffs-v0.2.c > /dev/null +// real 0m0.641s +// user 0m0.585s +// sys 0m0.037s // // More commentary is at: // https://godoc.org/github.com/google/wuffs/lib/dumbindent diff --git a/cmd/ractool/main.go b/cmd/ractool/main.go index 6aaca3b9..92f5968d 100644 --- a/cmd/ractool/main.go +++ b/cmd/ractool/main.go @@ -33,7 +33,7 @@ https://github.com/google/wuffs/blob/main/doc/spec/rac-spec.md Usage: - ractool [flags] [input_filename] + ractool [flags] [input_filename] If no input_filename is given, stdin is used. Either way, output is written to stdout. @@ -63,10 +63,10 @@ or temporary disk space. Examples: - ractool -decode foo.rac | sha256sum - ractool -decode -drange=400..500 foo.rac - ractool -encode foo.dat > foo.rac - ractool -encode -codec=zlib -dchunksize=256k foo.dat > foo.rac + ractool -decode foo.rac | sha256sum + ractool -decode -drange=400..500 foo.rac + ractool -encode foo.dat > foo.rac + ractool -encode -codec=zlib -dchunksize=256k foo.dat > foo.rac The "400..500" flag value means the 100 bytes ranging from a DSpace offset (offset in terms of decompressed bytes, not compressed bytes) of 400 @@ -79,42 +79,42 @@ Similarly, "1m" and "1M" both mean 1 mebibyte (1048576 bytes). General Flags: - -decode - whether to decode the input - -encode - whether to encode the input - -quiet - whether to suppress messages + -decode + whether to decode the input + -encode + whether to encode the input + -quiet + whether to suppress messages Decode-Related Flags: - -drange - the "i..j" range to decompress, "..8" means the first 8 bytes - -singlethreaded - whether to decode on a single execution thread + -drange + the "i..j" range to decompress, "..8" means the first 8 bytes + -singlethreaded + whether to decode on a single execution thread Encode-Related Flags: - -cchunksize - the chunk size (in CSpace) - -codec - the compression codec (default "zstd") - -cpagesize - the page size (in CSpace) - -dchunksize - the chunk size (in DSpace) - -indexlocation - the index location, "start" or "end" (default "start") - -resources - comma-separated list of resource files, such as shared dictionaries - -tmpdir - directory (e.g. $TMPDIR) for intermediate work; empty means in-memory + -cchunksize + the chunk size (in CSpace) + -codec + the compression codec (default "zstd") + -cpagesize + the page size (in CSpace) + -dchunksize + the chunk size (in DSpace) + -indexlocation + the index location, "start" or "end" (default "start") + -resources + comma-separated list of resource files, such as shared dictionaries + -tmpdir + directory (e.g. $TMPDIR) for intermediate work; empty means in-memory Codecs: - lz4 - zlib - zstd + lz4 + zlib + zstd Only zlib is fully supported. The others will work for the flags' default values, but they (1) don't support -cchunksize, only -dchunksize, and (2) don't @@ -125,123 +125,123 @@ Installation: Like any other implemented-in-Go program, to install the ractool program: - go install github.com/google/wuffs/cmd/ractool + go install github.com/google/wuffs/cmd/ractool Extended Example: - -------- - $ # Fetch and unzip the enwik8 test file, a sample of Wikipedia. - $ wget http://mattmahoney.net/dc/enwik8.zip - $ unzip enwik8.zip + -------- + $ # Fetch and unzip the enwik8 test file, a sample of Wikipedia. + $ wget http://mattmahoney.net/dc/enwik8.zip + $ unzip enwik8.zip - $ # Also zstd-encode it, as a reference point. Using compression level 15, - $ # instead of the default of 3, matches what ractool uses. - $ zstd -15 enwik8 + $ # Also zstd-encode it, as a reference point. Using compression level 15, + $ # instead of the default of 3, matches what ractool uses. + $ zstd -15 enwik8 - $ # Create a shared dictionary. Using zstd-the-program produces a - $ # dictionary that is especially useful for zstd-the-format, but it can - $ # also be used by other formats as a 'raw' prefix dictionary. - $ zstd -15 --train -B64K --maxdict=32K -o dict.dat enwik8 + $ # Create a shared dictionary. Using zstd-the-program produces a + $ # dictionary that is especially useful for zstd-the-format, but it can + $ # also be used by other formats as a 'raw' prefix dictionary. + $ zstd -15 --train -B64K --maxdict=32K -o dict.dat enwik8 - $ # RAC-encode it with various codecs, with and without that dictionary. - $ ractool -encode -codec=zlib -resources=dict.dat enwik8 > zlib.withdict.rac - $ ractool -encode -codec=zlib enwik8 > zlib.sansdict.rac - $ ractool -encode -codec=zstd -resources=dict.dat enwik8 > zstd.withdict.rac - $ ractool -encode -codec=zstd enwik8 > zstd.sansdict.rac - $ ractool -encode -codec=lz4 enwik8 > lz4.sansdict.rac + $ # RAC-encode it with various codecs, with and without that dictionary. + $ ractool -encode -codec=zlib -resources=dict.dat enwik8 > zlib.withdict.rac + $ ractool -encode -codec=zlib enwik8 > zlib.sansdict.rac + $ ractool -encode -codec=zstd -resources=dict.dat enwik8 > zstd.withdict.rac + $ ractool -encode -codec=zstd enwik8 > zstd.sansdict.rac + $ ractool -encode -codec=lz4 enwik8 > lz4.sansdict.rac - $ # The size overhead (comparing RAC+Xxx to Xxx) is about 0.2% (with) or - $ # 4.8% (sans) for zlib/zip and about 13% (with) or 28% (sans) for zstd, - $ # depending on whether we used a shared dictionary (with or sans). - $ ls -l - total 362080 - -rw-r----- 1 tao tao 32768 Oct 25 10:10 dict.dat - -rw-r----- 1 tao tao 100000000 Jun 2 2011 enwik8 - -rw-r----- 1 tao tao 36445475 Sep 2 2011 enwik8.zip - -rw-r----- 1 tao tao 29563109 Jun 2 2011 enwik8.zst - -rw-r----- 1 tao tao 58813316 Oct 25 10:17 lz4.sansdict.rac - -rw-r----- 1 tao tao 38185178 Oct 25 10:16 zlib.sansdict.rac - -rw-r----- 1 tao tao 36505786 Oct 25 10:16 zlib.withdict.rac - -rw-r----- 1 tao tao 37820491 Oct 25 10:17 zstd.sansdict.rac - -rw-r----- 1 tao tao 33386395 Oct 25 10:17 zstd.withdict.rac + $ # The size overhead (comparing RAC+Xxx to Xxx) is about 0.2% (with) or + $ # 4.8% (sans) for zlib/zip and about 13% (with) or 28% (sans) for zstd, + $ # depending on whether we used a shared dictionary (with or sans). + $ ls -l + total 362080 + -rw-r----- 1 tao tao 32768 Oct 25 10:10 dict.dat + -rw-r----- 1 tao tao 100000000 Jun 2 2011 enwik8 + -rw-r----- 1 tao tao 36445475 Sep 2 2011 enwik8.zip + -rw-r----- 1 tao tao 29563109 Jun 2 2011 enwik8.zst + -rw-r----- 1 tao tao 58813316 Oct 25 10:17 lz4.sansdict.rac + -rw-r----- 1 tao tao 38185178 Oct 25 10:16 zlib.sansdict.rac + -rw-r----- 1 tao tao 36505786 Oct 25 10:16 zlib.withdict.rac + -rw-r----- 1 tao tao 37820491 Oct 25 10:17 zstd.sansdict.rac + -rw-r----- 1 tao tao 33386395 Oct 25 10:17 zstd.withdict.rac - $ # Check that the decompressed forms all match. - $ cat enwik8 | sha256sum - 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - - $ unzip -p enwik8.zip | sha256sum - 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - - $ unzstd --stdout enwik8.zst | sha256sum - 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - - $ for f in *.rac; do ractool -decode $f | sha256sum; done - 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - - 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - - 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - - 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - - 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - + $ # Check that the decompressed forms all match. + $ cat enwik8 | sha256sum + 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - + $ unzip -p enwik8.zip | sha256sum + 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - + $ unzstd --stdout enwik8.zst | sha256sum + 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - + $ for f in *.rac; do ractool -decode $f | sha256sum; done + 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - + 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - + 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - + 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - + 2b49720ec4d78c3c9fabaee6e4179a5e997302b3a70029f30f2d582218c024a8 - - $ # Compare how long it takes to produce 8 bytes from the middle of - $ # the decompressed file, which happens to be the word "Business". - $ time unzip -p enwik8.zip | dd if=/dev/stdin status=none \ - > iflag=skip_bytes,count_bytes skip=50000000 count=8 - Business - real 0m0.379s - user 0m0.410s - sys 0m0.080s - $ time unzstd --stdout enwik8.zst | dd if=/dev/stdin status=none \ - > iflag=skip_bytes,count_bytes skip=50000000 count=8 - Business - real 0m0.172s - user 0m0.141s - sys 0m0.103s - $ time ractool -decode -drange=50000000..50000008 zstd.withdict.rac - Business - real 0m0.004s - user 0m0.005s - sys 0m0.001s + $ # Compare how long it takes to produce 8 bytes from the middle of + $ # the decompressed file, which happens to be the word "Business". + $ time unzip -p enwik8.zip | dd if=/dev/stdin status=none \ + > iflag=skip_bytes,count_bytes skip=50000000 count=8 + Business + real 0m0.379s + user 0m0.410s + sys 0m0.080s + $ time unzstd --stdout enwik8.zst | dd if=/dev/stdin status=none \ + > iflag=skip_bytes,count_bytes skip=50000000 count=8 + Business + real 0m0.172s + user 0m0.141s + sys 0m0.103s + $ time ractool -decode -drange=50000000..50000008 zstd.withdict.rac + Business + real 0m0.004s + user 0m0.005s + sys 0m0.001s - $ # A RAC file's chunks can be decoded in parallel, unlike ZIP, - $ # substantially reducing the real (wall clock) time taken even - $ # though both of these files use DEFLATE (RFC 1951) compression. - $ # - $ # Comparing the -singlethreaded time suggests that zlib-the-library's - $ # DEFLATE implementation is faster than unzip's. - $ time unzip -p enwik8.zip > /dev/null - real 0m0.711s - user 0m0.690s - sys 0m0.021s - $ time ractool -decode -singlethreaded zlib.withdict.rac > /dev/null - real 0m0.519s - user 0m0.513s - sys 0m0.017s - $ time ractool -decode zlib.withdict.rac > /dev/null - real 0m0.052s - user 0m0.678s - sys 0m0.036s + $ # A RAC file's chunks can be decoded in parallel, unlike ZIP, + $ # substantially reducing the real (wall clock) time taken even + $ # though both of these files use DEFLATE (RFC 1951) compression. + $ # + $ # Comparing the -singlethreaded time suggests that zlib-the-library's + $ # DEFLATE implementation is faster than unzip's. + $ time unzip -p enwik8.zip > /dev/null + real 0m0.711s + user 0m0.690s + sys 0m0.021s + $ time ractool -decode -singlethreaded zlib.withdict.rac > /dev/null + real 0m0.519s + user 0m0.513s + sys 0m0.017s + $ time ractool -decode zlib.withdict.rac > /dev/null + real 0m0.052s + user 0m0.678s + sys 0m0.036s - $ # A similar comparison can be made for Zstandard. - $ time unzstd --stdout enwik8.zst > /dev/null - real 0m0.203s - user 0m0.187s - sys 0m0.016s - $ time ractool -decode -singlethreaded zstd.withdict.rac > /dev/null - real 0m0.235s - user 0m0.206s - sys 0m0.033s - $ time ractool -decode zstd.withdict.rac > /dev/null - real 0m0.037s - user 0m0.374s - sys 0m0.080s + $ # A similar comparison can be made for Zstandard. + $ time unzstd --stdout enwik8.zst > /dev/null + real 0m0.203s + user 0m0.187s + sys 0m0.016s + $ time ractool -decode -singlethreaded zstd.withdict.rac > /dev/null + real 0m0.235s + user 0m0.206s + sys 0m0.033s + $ time ractool -decode zstd.withdict.rac > /dev/null + real 0m0.037s + user 0m0.374s + sys 0m0.080s - $ # For reference, LZ4 numbers. - $ time ractool -decode -singlethreaded lz4.sansdict.rac > /dev/null - real 0m0.072s - user 0m0.053s - sys 0m0.021s - $ time ractool -decode lz4.sansdict.rac > /dev/null - real 0m0.024s - user 0m0.097s - sys 0m0.034s - -------- + $ # For reference, LZ4 numbers. + $ time ractool -decode -singlethreaded lz4.sansdict.rac > /dev/null + real 0m0.072s + user 0m0.053s + sys 0m0.021s + $ time ractool -decode lz4.sansdict.rac > /dev/null + real 0m0.024s + user 0m0.097s + sys 0m0.034s + -------- */ package main diff --git a/lang/ast/ast.go b/lang/ast/ast.go index d5655e4b..8adbfb4c 100644 --- a/lang/ast/ast.go +++ b/lang/ast/ast.go @@ -328,12 +328,12 @@ func (n *Raw) SetPackage(tm *t.Map, pkg t.ID) error { const MaxExprDepth = 255 // Expr is an expression, such as "i", "+j" or "k + l[m(n, o)].p": -// - ID0: <0|operator|IDOpenParen|IDOpenBracket|IDDotDot|IDDot> -// - ID2: <0|literal|ident> -// - LHS: -// - MHS: -// - RHS: -// - List0: function call args, assoc. op args or list members. +// - ID0: <0|operator|IDOpenParen|IDOpenBracket|IDDotDot|IDDot> +// - ID2: <0|literal|ident> +// - LHS: +// - MHS: +// - RHS: +// - List0: function call args, assoc. op args or list members. // // A zero ID0 means an identifier or literal in ID2, like `foo`, `42` or a // status literal like `"#foo"`. @@ -468,10 +468,10 @@ func NewExpr(flags Flags, operator t.ID, ident t.ID, lhs *Node, mhs *Node, rhs * // Assert is "assert RHS via ID2(args)", "choose etc", "pre etc", "inv etc" or // "post etc": -// - ID0: -// - ID2: <"-string literal> reason -// - RHS: -// - List0: reason arguments +// - ID0: +// - ID2: <"-string literal> reason +// - RHS: +// - List0: reason arguments type Assert Node func (n *Assert) AsNode() *Node { return (*Node)(n) } @@ -513,8 +513,8 @@ func NewAssert(keyword t.ID, condition *Expr, reason t.ID, args []*Node) *Assert } // Arg is "name:value". -// - ID2: name -// - RHS: value +// - ID2: name +// - RHS: value type Arg Node func (n *Arg) AsNode() *Node { return (*Node)(n) } @@ -530,9 +530,9 @@ func NewArg(name t.ID, value *Expr) *Arg { } // Assign is "LHS = RHS" or "LHS op= RHS" or "RHS": -// - ID0: operator -// - LHS: -// - RHS: +// - ID0: operator +// - LHS: +// - RHS: type Assign Node func (n *Assign) AsNode() *Node { return (*Node)(n) } @@ -550,8 +550,8 @@ func NewAssign(operator t.ID, lhs *Expr, rhs *Expr) *Assign { } // Var is "var ID2 LHS": -// - ID2: name -// - LHS: +// - ID2: name +// - LHS: type Var Node func (n *Var) AsNode() *Node { return (*Node)(n) } @@ -569,9 +569,9 @@ func NewVar(name t.ID, xType *TypeExpr) *Var { } // Field is a "name : type" struct field: -// - FlagsPrivateData is the initializer need not explicitly memset to zero. -// - ID2: name -// - LHS: +// - FlagsPrivateData is the initializer need not explicitly memset to zero. +// - ID2: name +// - LHS: type Field Node func (n *Field) AsNode() *Node { return (*Node)(n) } @@ -590,11 +590,11 @@ func NewField(flags Flags, name t.ID, xType *TypeExpr) *Field { // IOManip is "io_bind (io:LHS, data:MHS, history_position:RHS) { List2 }" or // "io_limit (io:LHS, limit:MHS) { List2 }": -// - ID0: -// - LHS: -// - MHS: -// - RHS: -// - List2: body +// - ID0: +// - LHS: +// - MHS: +// - RHS: +// - List2: body type IOManip Node func (n *IOManip) AsNode() *Node { return (*Node)(n) } @@ -617,16 +617,16 @@ func NewIOManip(keyword t.ID, io *Expr, arg1 *Expr, historyPosition *Expr, body // Iterate is // "iterate.ID1 (assigns)(length:ID2, advance:ID0, unroll:LHS), List1 { List2 } else RHS": -// - FlagsHasBreak is the iterate has an explicit break -// - FlagsHasContinue is the iterate has an explicit continue -// - ID0: advance -// - ID1: <0|label> -// - ID2: length -// - LHS: unroll -// - RHS: -// - List0: assigns -// - List1: asserts -// - List2: body +// - FlagsHasBreak is the iterate has an explicit break +// - FlagsHasContinue is the iterate has an explicit continue +// - ID0: advance +// - ID1: <0|label> +// - ID2: length +// - LHS: unroll +// - RHS: +// - List0: assigns +// - List1: asserts +// - List2: body type Iterate Node func (n *Iterate) AsNode() *Node { return (*Node)(n) } @@ -661,12 +661,12 @@ func NewIterate(label t.ID, assigns []*Node, length t.ID, advance t.ID, unroll t } // While is "while.ID1 MHS, List1 { List2 } endwhile.ID1": -// - FlagsHasBreak is the while has an explicit break -// - FlagsHasContinue is the while has an explicit continue -// - ID1: <0|label> -// - MHS: -// - List1: asserts -// - List2: body +// - FlagsHasBreak is the while has an explicit break +// - FlagsHasContinue is the while has an explicit continue +// - ID1: <0|label> +// - MHS: +// - List1: asserts +// - List2: body // // TODO: should we be able to unroll while loops too? type While Node @@ -699,10 +699,10 @@ func NewWhile(label t.ID, condition *Expr, asserts []*Node) *While { } // If is "if MHS { List2 } else RHS" or "if MHS { List2 } else { List1 }": -// - MHS: -// - RHS: -// - List1: if-false body -// - List2: if-true body +// - MHS: +// - RHS: +// - List1: if-false body +// - List2: if-true body type If Node func (n *If) AsNode() *Node { return (*Node)(n) } @@ -722,8 +722,8 @@ func NewIf(condition *Expr, bodyIfTrue []*Node, bodyIfFalse []*Node, elseIf *If) } // Choose is "choose ID2: List0": -// - ID2: name -// - List0: method names. +// - ID2: name +// - List0: method names. type Choose Node func (n *Choose) AsNode() *Node { return (*Node)(n) } @@ -739,9 +739,9 @@ func NewChoose(name t.ID, args []*Node) *Choose { } // Ret is "return LHS" or "yield LHS": -// - FlagsReturnsError LHS is an error status -// - ID0: -// - LHS: +// - FlagsReturnsError LHS is an error status +// - ID0: +// - LHS: type Ret Node func (n *Ret) AsNode() *Node { return (*Node)(n) } @@ -760,8 +760,8 @@ func NewRet(keyword t.ID, value *Expr) *Ret { } // Jump is "break" or "continue", with an optional label, "break.label": -// - ID0: -// - ID1: <0|label> +// - ID0: +// - ID1: <0|label> type Jump Node func (n *Jump) AsNode() *Node { return (*Node)(n) } @@ -784,12 +784,12 @@ const MaxTypeExprDepth = 63 // TypeExpr is a type expression, such as "base.u32", "base.u32[..= 8]", "foo", // "pkg.bar", "ptr T", "array[8] T", "slice T" or "table T": -// - ID0: <0|IDArray|IDFunc|IDNptr|IDPtr|IDSlice|IDTable> -// - ID1: <0|pkg> -// - ID2: <0|type name> -// - LHS: -// - MHS: -// - RHS: +// - ID0: <0|IDArray|IDFunc|IDNptr|IDPtr|IDSlice|IDTable> +// - ID1: <0|pkg> +// - ID2: <0|type name> +// - LHS: +// - MHS: +// - RHS: // // An IDNptr or IDPtr ID0 means "nptr RHS" or "ptr RHS". RHS is the inner type. // @@ -958,14 +958,14 @@ func NewTypeExpr(decorator t.ID, pkg t.ID, name t.ID, alenRecvMin *Node, max *Ex const MaxBodyDepth = 255 // Func is "func ID2.ID0(LHS)(RHS) { List2 }": -// - FlagsPublic is "pub" vs "pri" -// - ID0: funcName -// - ID1: <0|receiverPkg> (set by calling SetPackage) -// - ID2: <0|receiverName> -// - LHS: in-parameters -// - RHS: out-parameters -// - List1: asserts -// - List2: body +// - FlagsPublic is "pub" vs "pri" +// - ID0: funcName +// - ID1: <0|receiverPkg> (set by calling SetPackage) +// - ID2: <0|receiverName> +// - LHS: in-parameters +// - RHS: out-parameters +// - List1: asserts +// - List2: body type Func Node func (n *Func) AsNode() *Node { return (*Node)(n) } @@ -1033,9 +1033,9 @@ func NewFunc(flags Flags, filename string, line uint32, receiverName t.ID, funcN } // Status is "error (RHS) ID2" or "suspension (RHS) ID2": -// - FlagsPublic is "pub" vs "pri" -// - ID1: <0|pkg> (set by calling SetPackage) -// - ID2: message +// - FlagsPublic is "pub" vs "pri" +// - ID1: <0|pkg> (set by calling SetPackage) +// - ID2: message type Status Node func (n *Status) AsNode() *Node { return (*Node)(n) } @@ -1055,11 +1055,11 @@ func NewStatus(flags Flags, filename string, line uint32, message t.ID) *Status } // Const is "const ID2 LHS = RHS": -// - FlagsPublic is "pub" vs "pri" -// - ID1: <0|pkg> (set by calling SetPackage) -// - ID2: name -// - LHS: -// - RHS: +// - FlagsPublic is "pub" vs "pri" +// - ID1: <0|pkg> (set by calling SetPackage) +// - ID2: name +// - LHS: +// - RHS: type Const Node func (n *Const) AsNode() *Node { return (*Node)(n) } @@ -1087,12 +1087,12 @@ func NewConst(flags Flags, filename string, line uint32, name t.ID, xType *TypeE const MaxImplements = 63 // Struct is "struct ID2? implements List0 (List1)": -// - FlagsPublic is "pub" vs "pri" -// - FlagsClassy is "ID2" vs "ID2?" -// - ID1: <0|pkg> (set by calling SetPackage) -// - ID2: name -// - List0: implements -// - List1: fields +// - FlagsPublic is "pub" vs "pri" +// - FlagsClassy is "ID2" vs "ID2?" +// - ID1: <0|pkg> (set by calling SetPackage) +// - ID2: name +// - List0: implements +// - List1: fields // // The question mark indicates a classy struct - one that supports methods, // especially coroutines. @@ -1120,7 +1120,7 @@ func NewStruct(flags Flags, filename string, line uint32, name t.ID, implements } // Use is "use ID2": -// - ID2: <"-string literal> package path +// - ID2: <"-string literal> package path type Use Node func (n *Use) AsNode() *Node { return (*Node)(n) } @@ -1138,7 +1138,7 @@ func NewUse(filename string, line uint32, path t.ID) *Use { } // File is a file of source code: -// - List0: top-level declarations +// - List0: top-level declarations type File Node func (n *File) AsNode() *Node { return (*Node)(n) } diff --git a/lang/token/list.go b/lang/token/list.go index acd6e50b..a4a83f7e 100644 --- a/lang/token/list.go +++ b/lang/token/list.go @@ -201,16 +201,16 @@ type Token struct { } // nBuiltInIDs is the number of built-in IDs. The packing is: -// - 0x00 is invalid. -// - 0x01 ..= 0x0F are squiggly punctuation, such as ";", "." and "?". -// - 0x10 ..= 0x1F are squiggly bookends, such as "(", ")" and "]". -// - 0x20 ..= 0x3F are squiggly assignments, such as "=" and "+=". -// - 0x40 ..= 0x6F are operators, such as "+", "==" and "not". -// - 0x70 ..= 0xAF are x-ops (disambiguation forms): unary vs binary "+". -// - 0xB0 ..= 0xCF are keywords, such as "if" and "return". -// - 0xD0 ..= 0xDF are type modifiers, such as "ptr" and "slice". -// - 0xE0 ..= 0xFF are literals, such as "ok" and "true". -// - 0x100 ..= 0x3FF are identifiers, such as "bool", "u32" and "read_u8". +// - 0x00 is invalid. +// - 0x01 ..= 0x0F are squiggly punctuation, such as ";", "." and "?". +// - 0x10 ..= 0x1F are squiggly bookends, such as "(", ")" and "]". +// - 0x20 ..= 0x3F are squiggly assignments, such as "=" and "+=". +// - 0x40 ..= 0x6F are operators, such as "+", "==" and "not". +// - 0x70 ..= 0xAF are x-ops (disambiguation forms): unary vs binary "+". +// - 0xB0 ..= 0xCF are keywords, such as "if" and "return". +// - 0xD0 ..= 0xDF are type modifiers, such as "ptr" and "slice". +// - 0xE0 ..= 0xFF are literals, such as "ok" and "true". +// - 0x100 ..= 0x3FF are identifiers, such as "bool", "u32" and "read_u8". // // Squiggly means a sequence of non-alpha-numeric bytes, such as "+" and "&=". const ( @@ -1299,9 +1299,12 @@ func init() { // addXForms modifies table so that, if table[x] == y, then table[y] = y. // // For example, for the unaryForms table, the explicit entries are like: -// IDPlus: IDXUnaryPlus, +// +// IDPlus: IDXUnaryPlus, +// // and this function implicitly addes entries like: -// IDXUnaryPlus: IDXUnaryPlus, +// +// IDXUnaryPlus: IDXUnaryPlus, func addXForms(table *[nBuiltInSymbolicIDs]ID) { implicitEntries := [nBuiltInSymbolicIDs]bool{} for _, y := range table { diff --git a/lib/dumbindent/dumbindent.go b/lib/dumbindent/dumbindent.go index c73f7577..1881df5d 100644 --- a/lib/dumbindent/dumbindent.go +++ b/lib/dumbindent/dumbindent.go @@ -27,57 +27,57 @@ // repository was 80 times faster than `clang-format`, even without a column // limit: // -// $ wc release/c/wuffs-v0.2.c -// 11858 35980 431885 release/c/wuffs-v0.2.c -// $ time dumbindent < release/c/wuffs-v0.2.c > /dev/null -// real 0m0.008s -// user 0m0.005s -// sys 0m0.005s -// $ time clang-format-9 < release/c/wuffs-v0.2.c > /dev/null -// real 0m0.668s -// user 0m0.618s -// sys 0m0.032s -// $ time clang-format-9 -style='{ColumnLimit: 0}' < release/c/wuffs-v0.2.c > /dev/null -// real 0m0.641s -// user 0m0.585s -// sys 0m0.037s +// $ wc release/c/wuffs-v0.2.c +// 11858 35980 431885 release/c/wuffs-v0.2.c +// $ time dumbindent < release/c/wuffs-v0.2.c > /dev/null +// real 0m0.008s +// user 0m0.005s +// sys 0m0.005s +// $ time clang-format-9 < release/c/wuffs-v0.2.c > /dev/null +// real 0m0.668s +// user 0m0.618s +// sys 0m0.032s +// $ time clang-format-9 -style='{ColumnLimit: 0}' < release/c/wuffs-v0.2.c > /dev/null +// real 0m0.641s +// user 0m0.585s +// sys 0m0.037s // // Apart from some rare and largely uninteresting exceptions, the dumbindent // algorithm only considers: // -// ∙ '{' and '}' curly braces, -// ∙ '(' and ')' round parentheses, -// ∙ '\n' line breaks, -// ∙ ' ' spaces and '\t' tabs that start or end a line, and -// ∙ strings, comments and preprocessor directives (in order to ignore any of -// the above special characters within them), +// ∙ '{' and '}' curly braces, +// ∙ '(' and ')' round parentheses, +// ∙ '\n' line breaks, +// ∙ ' ' spaces and '\t' tabs that start or end a line, and +// ∙ strings, comments and preprocessor directives (in order to ignore any of +// the above special characters within them), // // Everything else is an opaque byte. Consider this input: // -// for (i = 0; i < 3; i++) { -// j = 0; // Ignore { in a comment. -// if (i < j) { foo(); } -// u = (v + -// w); -// } +// for (i = 0; i < 3; i++) { +// j = 0; // Ignore { in a comment. +// if (i < j) { foo(); } +// u = (v + +// w); +// } // // From the algorithm's point of view, this input is equivalent to: // -// ....(.................).{ -// ................................. -// ...(.....).{....()..} -// ....(... -// .); -// } +// ....(.................).{ +// ................................. +// ...(.....).{....()..} +// ....(... +// .); +// } // // The formatted output (using the default of 2 spaces per indent level) is: // -// ....(.................).{ -// ................................. -// ...(.....).{....()..} -// ....(... -// .); -// } +// ....(.................).{ +// ................................. +// ...(.....).{....()..} +// ....(... +// .); +// } // // Dumbindent adjusts lines horizontally (indenting) but not vertically (it // does not break or un-break lines, or collapse consecutive blank lines), @@ -103,9 +103,9 @@ // // Similarly, dumbindent will not correct this mis-indentation: // -// if (condition) -// goto fail; -// goto fail; +// if (condition) +// goto fail; +// goto fail; // // Instead, when automatically or manually generating the input for dumbindent, // it is recommended to always emit curly braces (again, similar to `gofmt` diff --git a/lib/flatecut/flatecut.go b/lib/flatecut/flatecut.go index 0557168a..4515c5dd 100644 --- a/lib/flatecut/flatecut.go +++ b/lib/flatecut/flatecut.go @@ -130,23 +130,23 @@ func (b *bitstream) take(nBits uint32) int32 { // 4, 4)" example from RFC 1951 section 3.2.2, the huffman struct is // initialized by calling: // -// h.construct([]uint32{ -// 'A': 3, 'B': 3, 'C': 3, 'D': 3, 'E': 3, 'F': 2, 'G': 4, 'H': 4, -// }) +// h.construct([]uint32{ +// 'A': 3, 'B': 3, 'C': 3, 'D': 3, 'E': 3, 'F': 2, 'G': 4, 'H': 4, +// }) // // which results in: // -// huffman{ -// counts: [maxCodeBits + 1]uint32{ -// 2: 1, 3: 5, 4: 2, -// }, -// symbols: [maxNumCodes]int32{ -// 0: 'F', 1: 'A', 2: 'B', 3: 'C', 4: 'D', 5: 'E', 6: 'G', 7: 'H', -// }, -// lookUpTable: [256]uint32{ -// etc, -// }, -// } +// huffman{ +// counts: [maxCodeBits + 1]uint32{ +// 2: 1, 3: 5, 4: 2, +// }, +// symbols: [maxNumCodes]int32{ +// 0: 'F', 1: 'A', 2: 'B', 3: 'C', 4: 'D', 5: 'E', 6: 'G', 7: 'H', +// }, +// lookUpTable: [256]uint32{ +// etc, +// }, +// } // // Continuing the example from the RFC, decoding "1110" from the bitstream will // produce the 'G' symbol. diff --git a/lib/interval/interval.go b/lib/interval/interval.go index b1d2ea44..d70d88cb 100644 --- a/lib/interval/interval.go +++ b/lib/interval/interval.go @@ -1057,18 +1057,19 @@ func orOneNegOneNonNeg(neg IntRange, non IntRange) (z IntRange) { // possible xx in x and yy in y. // // Algorithm: -// // If the two intervals overlap, the result is the minimum of the two -// // intervals' maxima. -// // -// // This overlaps code path is just an optimization. -// if overlaps(x, y) { -// return min(xMax, yMax) -// } -// xFlip = bitFillRight(bitFillRight(xMax & ~xMin) & xMax & ~yMax) -// xResult = yMax & ((xMax & ~xFlip) | (xFlip >> 1)) -// yFlip = bitFillRight(bitFillRight(yMax & ~yMin) & yMax & ~xMax) -// yResult = xMax & ((yMax & ~yFlip) | (yFlip >> 1)) -// return max(xResult, yResult) +// +// // If the two intervals overlap, the result is the minimum of the two +// // intervals' maxima. +// // +// // This overlaps code path is just an optimization. +// if overlaps(x, y) { +// return min(xMax, yMax) +// } +// xFlip = bitFillRight(bitFillRight(xMax & ~xMin) & xMax & ~yMax) +// xResult = yMax & ((xMax & ~xFlip) | (xFlip >> 1)) +// yFlip = bitFillRight(bitFillRight(yMax & ~yMin) & yMax & ~xMax) +// yResult = xMax & ((yMax & ~yFlip) | (yFlip >> 1)) +// return max(xResult, yResult) // // If xMin and yMin are both zero, the overlaps branch is taken. // @@ -1189,13 +1190,15 @@ func (x IntRange) andMax(y IntRange) *big.Int { // possible xx in x and yy in y. // // Algorithm: -// droppable = bitFillRight((xMax & ~xMin) | (yMax & ~yMin)) -// available = xMax & yMax & droppable -// return xMax | yMax | (bitFillRight(available) >> 1) +// +// droppable = bitFillRight((xMax & ~xMin) | (yMax & ~yMin)) +// available = xMax & yMax & droppable +// return xMax | yMax | (bitFillRight(available) >> 1) // // If xMin and yMin are both zero, this simplifies to: -// available = xMax & yMax -// return xMax | yMax | (bitFillRight(available) >> 1) +// +// available = xMax & yMax +// return xMax | yMax | (bitFillRight(available) >> 1) func (x IntRange) orMax(y IntRange) *big.Int { if x[0].Sign() == 0 && y[0].Sign() == 0 { i := big.NewInt(0) @@ -1257,17 +1260,17 @@ func (x IntRange) orMax(y IntRange) *big.Int { } // bitFillRight modifies i to round up to the next power of 2 minus 1: -// - If i is +0 then bitFillRight(i) sets i to 0. -// - If i is +1 then bitFillRight(i) sets i to 1. -// - If i is +2 then bitFillRight(i) sets i to 3. -// - If i is +3 then bitFillRight(i) sets i to 3. -// - If i is +4 then bitFillRight(i) sets i to 7. -// - If i is +5 then bitFillRight(i) sets i to 7. -// - If i is +6 then bitFillRight(i) sets i to 7. -// - If i is +7 then bitFillRight(i) sets i to 7. -// - If i is +8 then bitFillRight(i) sets i to 15. -// - If i is +9 then bitFillRight(i) sets i to 15. -// - Etc. +// - If i is +0 then bitFillRight(i) sets i to 0. +// - If i is +1 then bitFillRight(i) sets i to 1. +// - If i is +2 then bitFillRight(i) sets i to 3. +// - If i is +3 then bitFillRight(i) sets i to 3. +// - If i is +4 then bitFillRight(i) sets i to 7. +// - If i is +5 then bitFillRight(i) sets i to 7. +// - If i is +6 then bitFillRight(i) sets i to 7. +// - If i is +7 then bitFillRight(i) sets i to 7. +// - If i is +8 then bitFillRight(i) sets i to 15. +// - If i is +9 then bitFillRight(i) sets i to 15. +// - Etc. func bitFillRight(i *big.Int) { if s := i.Sign(); s < 0 { panic("pre-condition failure") diff --git a/lib/litonlylzma/litonlylzma.go b/lib/litonlylzma/litonlylzma.go index 7d4af85b..53ce400e 100644 --- a/lib/litonlylzma/litonlylzma.go +++ b/lib/litonlylzma/litonlylzma.go @@ -30,23 +30,23 @@ // // Example compression numbers on a small English text file (at // https://github.com/google/wuffs/blob/main/test/data/romeo.txt): -// - romeo.txt is 942 bytes (100%). -// - romeo.txt.litonlylzma is 659 bytes (70%). -// - romeo.txt.xz is 644 bytes (68%). -// - romeo.txt.lzma is 598 bytes (63%). -// - romeo.txt.bz2 is 568 bytes (60%). -// - romeo.txt.zst is 559 bytes (59%). -// - romeo.txt.gz is 558 bytes (59%). +// - romeo.txt is 942 bytes (100%). +// - romeo.txt.litonlylzma is 659 bytes (70%). +// - romeo.txt.xz is 644 bytes (68%). +// - romeo.txt.lzma is 598 bytes (63%). +// - romeo.txt.bz2 is 568 bytes (60%). +// - romeo.txt.zst is 559 bytes (59%). +// - romeo.txt.gz is 558 bytes (59%). // // Example compression numbers on a large archive of source code (at // https://cdn.kernel.org/pub/linux/kernel/v5.x/linux-5.0.1.tar.xz): -// - linux-5.0.1.tar is 863313920 bytes (100%). -// - linux-5.0.1.tar.litonlylzma is 449726070 bytes (52%). -// - linux-5.0.1.tar.gz is 164575959 bytes (19%). -// - linux-5.0.1.tar.zst is 156959897 bytes (18%). -// - linux-5.0.1.tar.bz2 is 125873134 bytes (15%). -// - linux-5.0.1.tar.xz is 108233572 bytes (13%). -// - linux-5.0.1.tar.lzma is 108216601 bytes (13%). +// - linux-5.0.1.tar is 863313920 bytes (100%). +// - linux-5.0.1.tar.litonlylzma is 449726070 bytes (52%). +// - linux-5.0.1.tar.gz is 164575959 bytes (19%). +// - linux-5.0.1.tar.zst is 156959897 bytes (18%). +// - linux-5.0.1.tar.bz2 is 125873134 bytes (15%). +// - linux-5.0.1.tar.xz is 108233572 bytes (13%). +// - linux-5.0.1.tar.lzma is 108216601 bytes (13%). // // The various tools (/usr/bin/gzip, /usr/bin/xz, etc) were all ran with their // default settings, not their "maximum compression" settings. This is why the @@ -260,39 +260,39 @@ func (p *prob) encodeBit(rEnc *rangeEncoder, bitValue uint32) { // byteProbs hold probabilities for coding 8 bits (1 byte) of data. It is a 256 // element array such that: -// - The 0th element is unused. -// - The 1st element holds the probability that the 7th bit (the highest, most -// significant bit) is 0. -// - The 2nd element holds the probability that the 6th bit is 0, conditional -// on the high bit being 0. -// - The 3rd element holds the probability that the 6th bit is 0, conditional -// on the high bit being 1. -// - The 4th element holds the probability that the 5th bit is 0, conditional -// on the high two bits being 00. -// - The 5th element holds the probability that the 5th bit is 0, conditional -// on the high two bits being 01. -// - The 6th element holds the probability that the 5th bit is 0, conditional -// on the high two bits being 10. -// - The 7th element holds the probability that the 5th bit is 0, conditional -// on the high two bits being 11. -// - The 8th element holds the probability that the 4th bit is 0, conditional -// on the high three bits being 000. -// - etc -// - The 255th element holds the probability that the 0th bit (the lowest, -// least significant bit) is 0, conditional on the high seven bits being -// 1111111. +// - The 0th element is unused. +// - The 1st element holds the probability that the 7th bit (the highest, most +// significant bit) is 0. +// - The 2nd element holds the probability that the 6th bit is 0, conditional +// on the high bit being 0. +// - The 3rd element holds the probability that the 6th bit is 0, conditional +// on the high bit being 1. +// - The 4th element holds the probability that the 5th bit is 0, conditional +// on the high two bits being 00. +// - The 5th element holds the probability that the 5th bit is 0, conditional +// on the high two bits being 01. +// - The 6th element holds the probability that the 5th bit is 0, conditional +// on the high two bits being 10. +// - The 7th element holds the probability that the 5th bit is 0, conditional +// on the high two bits being 11. +// - The 8th element holds the probability that the 4th bit is 0, conditional +// on the high three bits being 000. +// - etc +// - The 255th element holds the probability that the 0th bit (the lowest, +// least significant bit) is 0, conditional on the high seven bits being +// 1111111. // // Put another way, the 256 elements' value of N, as in "it's a probability for // the Nth bit", looks like this (when arranged in 8 rows of 32 elements): // -// u7665555444444443333333333333333 -// 22222222222222222222222222222222 -// 11111111111111111111111111111111 -// 11111111111111111111111111111111 -// 00000000000000000000000000000000 -// 00000000000000000000000000000000 -// 00000000000000000000000000000000 -// 00000000000000000000000000000000 +// u7665555444444443333333333333333 +// 22222222222222222222222222222222 +// 11111111111111111111111111111111 +// 11111111111111111111111111111111 +// 00000000000000000000000000000000 +// 00000000000000000000000000000000 +// 00000000000000000000000000000000 +// 00000000000000000000000000000000 // // The 'u' means that the 0th element is unused. type byteProbs [0x100]prob diff --git a/script/print-deflate-huff-table-size.go b/script/print-deflate-huff-table-size.go index 32cbe27f..d10f2c44 100644 --- a/script/print-deflate-huff-table-size.go +++ b/script/print-deflate-huff-table-size.go @@ -365,7 +365,7 @@ type bitVector []uint8 // zig packs a pair of non-negative integers (i, j) into a single unique // non-negative integer, in a zig-zagging pattern: // -// i=0 i=1 i=2 i=3 i=4 etc +// . i=0 i=1 i=2 i=3 i=4 etc // j=0: 0 1 3 6 10 ... // j=1: 2 4 7 11 16 ... // j=2: 5 8 12 17 23 ...