| 1 | ---
|
| 2 | default_highlighter: oils-sh
|
| 3 | ---
|
| 4 |
|
| 5 | YSH vs. Shell Idioms
|
| 6 | ====================
|
| 7 |
|
| 8 | This is an informal, lightly-organized list of recommended idioms for the
|
| 9 | [YSH]($xref) language. Each section has snippets labeled *No* and *Yes*.
|
| 10 |
|
| 11 | - Use the *Yes* style when you want to write in YSH, and don't care about
|
| 12 | compatibility with other shells.
|
| 13 | - The *No* style is discouraged in new code, but YSH will run it. The [OSH
|
| 14 | language]($xref:osh-language) is compatible with
|
| 15 | [POSIX]($xref:posix-shell-spec) and [bash]($xref).
|
| 16 |
|
| 17 | [J8 Notation]: j8-notation.html
|
| 18 |
|
| 19 | <!-- cmark.py expands this -->
|
| 20 | <div id="toc">
|
| 21 | </div>
|
| 22 |
|
| 23 | ## Use [Simple Word Evaluation](simple-word-eval.html) to Avoid "Quoting Hell"
|
| 24 |
|
| 25 | ### Substitute Variables
|
| 26 |
|
| 27 | No:
|
| 28 |
|
| 29 | local x='my song.mp3'
|
| 30 | ls "$x" # quotes required to avoid mangling
|
| 31 |
|
| 32 | Yes:
|
| 33 |
|
| 34 | var x = 'my song.mp3'
|
| 35 | ls $x # no quotes needed
|
| 36 |
|
| 37 | ### Splice Arrays
|
| 38 |
|
| 39 | No:
|
| 40 |
|
| 41 | local myflags=( --all --long )
|
| 42 | ls "${myflags[@]}" "$@"
|
| 43 |
|
| 44 | Yes:
|
| 45 |
|
| 46 | var myflags = :| --all --long |
|
| 47 | ls @myflags @ARGV
|
| 48 |
|
| 49 | ### Explicitly Split, Glob, and Omit Empty Args
|
| 50 |
|
| 51 | YSH doesn't split arguments after variable expansion.
|
| 52 |
|
| 53 | No:
|
| 54 |
|
| 55 | local packages='python-dev gawk'
|
| 56 | apt install $packages
|
| 57 |
|
| 58 | Yes:
|
| 59 |
|
| 60 | var packages = 'python-dev gawk'
|
| 61 | apt install @[split(packages)]
|
| 62 |
|
| 63 | Even better:
|
| 64 |
|
| 65 | var packages = :| python-dev gawk | # array literal
|
| 66 | apt install @packages # splice array
|
| 67 |
|
| 68 | ---
|
| 69 |
|
| 70 | YSH doesn't glob after variable expansion.
|
| 71 |
|
| 72 | No:
|
| 73 |
|
| 74 | local pat='*.py'
|
| 75 | echo $pat
|
| 76 |
|
| 77 |
|
| 78 | Yes:
|
| 79 |
|
| 80 | var pat = '*.py'
|
| 81 | echo @[glob(pat)] # explicit call
|
| 82 |
|
| 83 | ---
|
| 84 |
|
| 85 | YSH doesn't omit unquoted words that evaluate to the empty string.
|
| 86 |
|
| 87 | No:
|
| 88 |
|
| 89 | local e=''
|
| 90 | cp $e other $dest # cp gets 2 args, not 3, in sh
|
| 91 |
|
| 92 | Yes:
|
| 93 |
|
| 94 | var e = ''
|
| 95 | cp @[maybe(e)] other $dest # explicit call
|
| 96 |
|
| 97 | ### Iterate a Number of Times (Split Command Sub)
|
| 98 |
|
| 99 | No:
|
| 100 |
|
| 101 | local n=3
|
| 102 | for x in $(seq $n); do # No implicit splitting of unquoted words in YSH
|
| 103 | echo $x
|
| 104 | done
|
| 105 |
|
| 106 | OK:
|
| 107 |
|
| 108 | var n = 3
|
| 109 | for x in @(seq $n) { # Explicit splitting
|
| 110 | echo $x
|
| 111 | }
|
| 112 |
|
| 113 | Better;
|
| 114 |
|
| 115 | var n = 3
|
| 116 | for x in (1 .. n+1) { # Range, avoids external program
|
| 117 | echo $x
|
| 118 | }
|
| 119 |
|
| 120 | Note that `{1..3}` works in bash and YSH, but the numbers must be constant.
|
| 121 |
|
| 122 | ## Strings and Quoting
|
| 123 |
|
| 124 | ### Use J8 strings for C-style escapes, like `\n`
|
| 125 |
|
| 126 | Bash adds C-style strings to shell with the `$''` syntax. YSH looks more like
|
| 127 | Python or Rust:
|
| 128 |
|
| 129 | No:
|
| 130 |
|
| 131 | echo $'hi \n two' # two lines
|
| 132 |
|
| 133 | echo $'mu \u03bc -- smiley \U0001f642' # \u and \U escapes
|
| 134 |
|
| 135 | echo $'raw byte \xff'
|
| 136 |
|
| 137 | Yes:
|
| 138 |
|
| 139 | echo u'hi \n two' # two lines
|
| 140 |
|
| 141 | echo u'mu \u{3bc} -- smiley \u{1f642}' # one consistent syntax \u{}
|
| 142 |
|
| 143 | echo b'raw byte \yff' # byte escapes with \yff
|
| 144 |
|
| 145 | Note that the syntax of YSH **code** matches our syntax for **data**, which is
|
| 146 | based on JSON. See [Data Languages](ref/toc-data.html) > [J8
|
| 147 | Notation](ref/chap-j8.html).
|
| 148 |
|
| 149 | ### Use Multi-line Strings instead of Here Docs
|
| 150 |
|
| 151 | YSH has multi-line strings with `'''` and `"""`. Combined with the `<<<`
|
| 152 | operator, they replace here docs.
|
| 153 |
|
| 154 | No:
|
| 155 |
|
| 156 | cat <<EOF
|
| 157 | hello $name
|
| 158 | bye
|
| 159 | EOF
|
| 160 |
|
| 161 | Yes:
|
| 162 |
|
| 163 | cat <<< """
|
| 164 | hello $name
|
| 165 | bye
|
| 166 | """
|
| 167 |
|
| 168 | No:
|
| 169 |
|
| 170 | cat <<'EOF'
|
| 171 | prize is $4.99
|
| 172 | EOF
|
| 173 |
|
| 174 | Yes:
|
| 175 |
|
| 176 | cat <<< '''
|
| 177 | prize is $4.99
|
| 178 | '''
|
| 179 |
|
| 180 | If there's whitespace before the closing quote, the same whitespace is removed
|
| 181 | from every line. Compare with shell:
|
| 182 |
|
| 183 | No:
|
| 184 |
|
| 185 | # special <<- operator removes leading tabs
|
| 186 | cat <<-EOF
|
| 187 | hello $name
|
| 188 | EOF
|
| 189 | # ^^^^^^ tabs only, not spaces
|
| 190 |
|
| 191 | Yes:
|
| 192 |
|
| 193 | cat <<< """
|
| 194 | hello $name
|
| 195 | """
|
| 196 | # ^^ indent can consist of spaces, making code look nicer
|
| 197 |
|
| 198 | ## Avoid Ad Hoc Parsing and Splitting
|
| 199 |
|
| 200 | In other words, avoid *groveling through backslashes and spaces* in shell.
|
| 201 |
|
| 202 | Instead, emit and consume [J8 Notation]($xref:j8-notation):
|
| 203 |
|
| 204 | - J8 strings are [JSON]($xref) strings, with an upgrade for byte string
|
| 205 | literals
|
| 206 | - [JSON8]($xref) is [JSON]($xref), with this same upgrade
|
| 207 | - [TSV8]($xref) is TSV with this upgrade (not yet implemented)
|
| 208 |
|
| 209 | Custom parsing and serializing should be limited to "the edges" of your YSH
|
| 210 | programs.
|
| 211 |
|
| 212 | ### More Strategies For Structured Data
|
| 213 |
|
| 214 | - **Wrap** and Adapt External Tools. Parse their output, and emit [J8 Notation][].
|
| 215 | - These can be one-off, "bespoke" wrappers in your program, or maintained
|
| 216 | programs. Use the `proc` construct and `flagspec`!
|
| 217 | - Example: [uxy](https://github.com/sustrik/uxy) wrappers.
|
| 218 | - TODO: Examples written in YSH and in other languages.
|
| 219 | - **Patch** Existing Tools.
|
| 220 | - Enhance GNU grep, etc. to emit [J8 Notation][]. Add a
|
| 221 | `--j8` flag.
|
| 222 | - **Write Your Own** Structured Versions.
|
| 223 | - For example, you can write a structured subset of `ls` in Python with
|
| 224 | little effort.
|
| 225 |
|
| 226 | <!--
|
| 227 | ls -q and -Q already exist, but --j8 or --tsv8 is probably fine
|
| 228 | -->
|
| 229 |
|
| 230 | ## The `write` Builtin Is Simpler Than `printf` and `echo`
|
| 231 |
|
| 232 | ### Write an Arbitrary Line
|
| 233 |
|
| 234 | No:
|
| 235 |
|
| 236 | printf '%s\n' "$mystr"
|
| 237 |
|
| 238 | Yes:
|
| 239 |
|
| 240 | write -- $mystr
|
| 241 |
|
| 242 | The `write` builtin accepts `--` so it doesn't confuse flags and args.
|
| 243 |
|
| 244 | ### Write Without a Newline
|
| 245 |
|
| 246 | No:
|
| 247 |
|
| 248 | echo -n "$mystr" # breaks if mystr is -e
|
| 249 |
|
| 250 | Yes:
|
| 251 |
|
| 252 | write --end '' -- $mystr
|
| 253 | write -n -- $mystr # -n is an alias for --end ''
|
| 254 |
|
| 255 | ### Write an Array of Lines
|
| 256 |
|
| 257 | var myarray = :| one two three |
|
| 258 | write -- @myarray
|
| 259 |
|
| 260 | ## New Long Flags on the `read` builtin
|
| 261 |
|
| 262 | ### Read a Line
|
| 263 |
|
| 264 | No:
|
| 265 |
|
| 266 | read line # Mangles your backslashes!
|
| 267 |
|
| 268 | Better:
|
| 269 |
|
| 270 | read -r line # Still messes with leading and trailing whitespace
|
| 271 |
|
| 272 | IFS= read -r line # OK, but doesn't work in YSH
|
| 273 |
|
| 274 | Yes:
|
| 275 |
|
| 276 | read --raw-line # Gives you the line, without trailing \n
|
| 277 |
|
| 278 | (Note that `read --raw-line` is still an unbuffered read, which means it slowly
|
| 279 | reads a byte at a time. We plan to add buffered reads as well.)
|
| 280 |
|
| 281 | ### Read a Whole File
|
| 282 |
|
| 283 | No:
|
| 284 |
|
| 285 | read -d '' # harder to read, easy to forget -r
|
| 286 |
|
| 287 | Yes:
|
| 288 |
|
| 289 | read --all # sets $_reply
|
| 290 | read --all (&myvar) # sets $myvar
|
| 291 |
|
| 292 | ### Read Lines of a File
|
| 293 |
|
| 294 | No:
|
| 295 |
|
| 296 | # The IFS= idiom doesn't work in YSH, because of dynamic scope!
|
| 297 | while IFS= read -r line; do
|
| 298 | echo $line
|
| 299 | done
|
| 300 |
|
| 301 | Yes:
|
| 302 |
|
| 303 | while read --raw-line {
|
| 304 | echo $_reply
|
| 305 | }
|
| 306 | # this reads a byte at a time, unbuffered, like shell
|
| 307 |
|
| 308 | Yes:
|
| 309 |
|
| 310 | for line in (io.stdin) {
|
| 311 | echo $line
|
| 312 | }
|
| 313 | # this reads buffered lines, which is much faster
|
| 314 |
|
| 315 | ### Read a Number of Bytes
|
| 316 |
|
| 317 | No:
|
| 318 |
|
| 319 | read -n 3 # slow because it respects -d delim
|
| 320 | # also strips whitespace
|
| 321 |
|
| 322 | Better:
|
| 323 |
|
| 324 | read -N 3 # good behavior, but easily confused with -n
|
| 325 |
|
| 326 | Yes:
|
| 327 |
|
| 328 | read --num-bytes 3 # sets $_reply
|
| 329 | read --num-bytes 3 (&myvar) # sets $myvar
|
| 330 |
|
| 331 |
|
| 332 | ### Read Until `\0` (consume `find -print0`)
|
| 333 |
|
| 334 | No:
|
| 335 |
|
| 336 | # Obscure syntax that bash accepts, but not other shells
|
| 337 | read -r -d '' myvar
|
| 338 |
|
| 339 | Yes:
|
| 340 |
|
| 341 | read -0 (&myvar)
|
| 342 |
|
| 343 | ## YSH Enhancements to Builtins
|
| 344 |
|
| 345 | ### Use `shopt` Instead of `set`
|
| 346 |
|
| 347 | Using a single builtin for all options makes scripts easier to read:
|
| 348 |
|
| 349 | Discouraged:
|
| 350 |
|
| 351 | set -o errexit
|
| 352 | shopt -s dotglob
|
| 353 |
|
| 354 | Idiomatic:
|
| 355 |
|
| 356 | shopt --set errexit
|
| 357 | shopt --set dotglob
|
| 358 |
|
| 359 | (As always, `set` can be used when you care about compatibility with other
|
| 360 | shells.)
|
| 361 |
|
| 362 | ### Use `&` When Mentioning Variable Names
|
| 363 |
|
| 364 | YSH uses [places](variables.html#return-by-mutating-a-place-advanced) to make
|
| 365 | out-parameters of procs more explicit.
|
| 366 |
|
| 367 | No:
|
| 368 |
|
| 369 | read -0 record < file.bin
|
| 370 | echo $record
|
| 371 |
|
| 372 | Yes:
|
| 373 |
|
| 374 | read -0 (&myvar) < file.bin
|
| 375 | echo $record
|
| 376 |
|
| 377 |
|
| 378 | ### Consider Using `--long-flags`
|
| 379 |
|
| 380 | Easier to write:
|
| 381 |
|
| 382 | test -d /tmp
|
| 383 | test -d / && test -f /vmlinuz
|
| 384 |
|
| 385 | shopt -u extglob
|
| 386 |
|
| 387 | Easier to read:
|
| 388 |
|
| 389 | test --dir /tmp
|
| 390 | test --dir / && test --file /vmlinuz
|
| 391 |
|
| 392 | shopt --unset extglob
|
| 393 |
|
| 394 | ## Use Blocks to Save and Restore Context
|
| 395 |
|
| 396 | ### Do Something In Another Directory
|
| 397 |
|
| 398 | No:
|
| 399 |
|
| 400 | ( cd /tmp; echo $PWD ) # subshell is unnecessary (and limited)
|
| 401 |
|
| 402 | No:
|
| 403 |
|
| 404 | pushd /tmp
|
| 405 | echo $PWD
|
| 406 | popd
|
| 407 |
|
| 408 | Yes:
|
| 409 |
|
| 410 | cd /tmp {
|
| 411 | echo $PWD
|
| 412 | }
|
| 413 |
|
| 414 | ### Batch I/O
|
| 415 |
|
| 416 | No:
|
| 417 |
|
| 418 | echo 1 > out.txt
|
| 419 | echo 2 >> out.txt # appending is less efficient
|
| 420 | # because open() and close()
|
| 421 |
|
| 422 | No:
|
| 423 |
|
| 424 | { echo 1
|
| 425 | echo 2
|
| 426 | } > out.txt
|
| 427 |
|
| 428 | Yes:
|
| 429 |
|
| 430 | redir > out.txt {
|
| 431 | echo 1
|
| 432 | echo 2
|
| 433 | }
|
| 434 |
|
| 435 | The `redir` builtin is syntactic sugar -- it lets you see redirects before the
|
| 436 | code that uses them.
|
| 437 |
|
| 438 | ### Temporarily Set Shell Options
|
| 439 |
|
| 440 | No:
|
| 441 |
|
| 442 | set +o errexit
|
| 443 | myfunc # without error checking
|
| 444 | set -o errexit
|
| 445 |
|
| 446 | Yes:
|
| 447 |
|
| 448 | shopt --unset errexit {
|
| 449 | myfunc
|
| 450 | }
|
| 451 |
|
| 452 | ### Use the `forkwait` builtin for Subshells, not `()`
|
| 453 |
|
| 454 | No:
|
| 455 |
|
| 456 | ( cd /tmp; rm *.sh )
|
| 457 |
|
| 458 | Yes:
|
| 459 |
|
| 460 | forkwait {
|
| 461 | cd /tmp
|
| 462 | rm *.sh
|
| 463 | }
|
| 464 |
|
| 465 | Better:
|
| 466 |
|
| 467 | cd /tmp { # no process created
|
| 468 | rm *.sh
|
| 469 | }
|
| 470 |
|
| 471 | ### Use the `fork` builtin for async, not `&`
|
| 472 |
|
| 473 | No:
|
| 474 |
|
| 475 | myfunc &
|
| 476 |
|
| 477 | { sleep 1; echo one; sleep 2; } &
|
| 478 |
|
| 479 | Yes:
|
| 480 |
|
| 481 | fork { myfunc }
|
| 482 |
|
| 483 | fork { sleep 1; echo one; sleep 2 }
|
| 484 |
|
| 485 | ## Use Procs (Better Shell Functions)
|
| 486 |
|
| 487 | ### Use Named Parameters Instead of `$1`, `$2`, ...
|
| 488 |
|
| 489 | No:
|
| 490 |
|
| 491 | f() {
|
| 492 | local src=$1
|
| 493 | local dest=${2:-/tmp}
|
| 494 |
|
| 495 | cp "$src" "$dest"
|
| 496 | }
|
| 497 |
|
| 498 | Yes:
|
| 499 |
|
| 500 | proc f(src, dest='/tmp') { # Python-like default values
|
| 501 | cp $src $dest
|
| 502 | }
|
| 503 |
|
| 504 | ### Use Named Varargs Instead of `"$@"`
|
| 505 |
|
| 506 | No:
|
| 507 |
|
| 508 | f() {
|
| 509 | local first=$1
|
| 510 | shift
|
| 511 |
|
| 512 | echo $first
|
| 513 | echo "$@"
|
| 514 | }
|
| 515 |
|
| 516 | Yes:
|
| 517 |
|
| 518 | proc f(first, @rest) { # @ means "the rest of the arguments"
|
| 519 | write -- $first
|
| 520 | write -- @rest # @ means "splice this array"
|
| 521 | }
|
| 522 |
|
| 523 | You can also use the implicit `ARGV` variable:
|
| 524 |
|
| 525 | proc p {
|
| 526 | cp -- @ARGV /tmp
|
| 527 | }
|
| 528 |
|
| 529 | ### Use "Out Params" instead of `declare -n`
|
| 530 |
|
| 531 | Out params are one way to "return" values from a `proc`.
|
| 532 |
|
| 533 | No:
|
| 534 |
|
| 535 | f() {
|
| 536 | local in=$1
|
| 537 | local -n out=$2
|
| 538 |
|
| 539 | out=PREFIX-$in
|
| 540 | }
|
| 541 |
|
| 542 | myvar='init'
|
| 543 | f zzz myvar # assigns myvar to 'PREFIX-zzz'
|
| 544 |
|
| 545 |
|
| 546 | Yes:
|
| 547 |
|
| 548 | proc f(in, :out) { # : is an out param, i.e. a string "reference"
|
| 549 | setref out = "PREFIX-$in"
|
| 550 | }
|
| 551 |
|
| 552 | var myvar = 'init'
|
| 553 | f zzz :myvar # assigns myvar to 'PREFIX-zzz'.
|
| 554 | # colon is required
|
| 555 |
|
| 556 | ### Note: Procs Don't Mess With Their Callers
|
| 557 |
|
| 558 | That is, [dynamic scope]($xref:dynamic-scope) is turned off when procs are
|
| 559 | invoked.
|
| 560 |
|
| 561 | Here's an example of shell functions reading variables in their caller:
|
| 562 |
|
| 563 | bar() {
|
| 564 | echo $foo_var # looks up the stack
|
| 565 | }
|
| 566 |
|
| 567 | foo() {
|
| 568 | foo_var=x
|
| 569 | bar
|
| 570 | }
|
| 571 |
|
| 572 | foo
|
| 573 |
|
| 574 | In YSH, you have to pass params explicitly:
|
| 575 |
|
| 576 | proc bar {
|
| 577 | echo $foo_var # error, not defined
|
| 578 | }
|
| 579 |
|
| 580 | Shell functions can also **mutate** variables in their caller! But procs can't
|
| 581 | do this, which makes code easier to reason about.
|
| 582 |
|
| 583 | ## Use Modules
|
| 584 |
|
| 585 | YSH has a few lightweight features that make it easier to organize code into
|
| 586 | files. It doesn't have "namespaces".
|
| 587 |
|
| 588 | ### Relative Imports
|
| 589 |
|
| 590 | Suppose we are running `bin/mytool`, and we want `BASE_DIR` to be the root of
|
| 591 | the repository so we can do a relative import of `lib/foo.sh`.
|
| 592 |
|
| 593 | No:
|
| 594 |
|
| 595 | # All of these are common idioms, with caveats
|
| 596 | BASE_DIR=$(dirname $0)/..
|
| 597 |
|
| 598 | BASE_DIR=$(dirname ${BASH_SOURCE[0]})/..
|
| 599 |
|
| 600 | BASE_DIR=$(cd $($dirname $0)/.. && pwd)
|
| 601 |
|
| 602 | BASE_DIR=$(dirname (dirname $(readlink -f $0)))
|
| 603 |
|
| 604 | source $BASE_DIR/lib/foo.sh
|
| 605 |
|
| 606 | Yes:
|
| 607 |
|
| 608 | const BASE_DIR = "$this_dir/.."
|
| 609 |
|
| 610 | source $BASE_DIR/lib/foo.sh
|
| 611 |
|
| 612 | # Or simply:
|
| 613 | source $_this_dir/../lib/foo.sh
|
| 614 |
|
| 615 | The value of `_this_dir` is the directory that contains the currently executing
|
| 616 | file.
|
| 617 |
|
| 618 | ### Include Guards
|
| 619 |
|
| 620 | No:
|
| 621 |
|
| 622 | # libfoo.sh
|
| 623 | if test -z "$__LIBFOO_SH"; then
|
| 624 | return
|
| 625 | fi
|
| 626 | __LIBFOO_SH=1
|
| 627 |
|
| 628 | Yes:
|
| 629 |
|
| 630 | # libfoo.sh
|
| 631 | module libfoo.sh || return 0
|
| 632 |
|
| 633 | ### Taskfile Pattern
|
| 634 |
|
| 635 | No:
|
| 636 |
|
| 637 | deploy() {
|
| 638 | echo ...
|
| 639 | }
|
| 640 | "$@"
|
| 641 |
|
| 642 | Yes
|
| 643 |
|
| 644 | proc deploy() {
|
| 645 | echo ...
|
| 646 | }
|
| 647 | runproc @ARGV # gives better error messages
|
| 648 |
|
| 649 | ## Error Handling
|
| 650 |
|
| 651 | [YSH Fixes Shell's Error Handling (`errexit`)](error-handling.html) once and
|
| 652 | for all! Here's a comprehensive list of error handling idioms.
|
| 653 |
|
| 654 | ### Don't Use `&&` Outside of `if` / `while`
|
| 655 |
|
| 656 | It's implicit because `errexit` is on in YSH.
|
| 657 |
|
| 658 | No:
|
| 659 |
|
| 660 | mkdir /tmp/dest && cp foo /tmp/dest
|
| 661 |
|
| 662 | Yes:
|
| 663 |
|
| 664 | mkdir /tmp/dest
|
| 665 | cp foo /tmp/dest
|
| 666 |
|
| 667 | It also avoids the *Trailing `&&` Pitfall* mentioned at the end of the [error
|
| 668 | handling](error-handling.html) doc.
|
| 669 |
|
| 670 | ### Ignore an Error
|
| 671 |
|
| 672 | No:
|
| 673 |
|
| 674 | ls /bad || true # OK because ls is external
|
| 675 | myfunc || true # suffers from the "Disabled errexit Quirk"
|
| 676 |
|
| 677 | Yes:
|
| 678 |
|
| 679 | try { ls /bad }
|
| 680 | try { myfunc }
|
| 681 |
|
| 682 | ### Retrieve A Command's Status When `errexit` is On
|
| 683 |
|
| 684 | No:
|
| 685 |
|
| 686 | # set -e is enabled earlier
|
| 687 |
|
| 688 | set +e
|
| 689 | mycommand # this ignores errors when mycommand is a function
|
| 690 | status=$? # save it before it changes
|
| 691 | set -e
|
| 692 |
|
| 693 | echo $status
|
| 694 |
|
| 695 | Yes:
|
| 696 |
|
| 697 | try {
|
| 698 | mycommand
|
| 699 | }
|
| 700 | echo $[_error.code]
|
| 701 |
|
| 702 | ### Does a Builtin Or External Command Succeed?
|
| 703 |
|
| 704 | These idioms are OK in both shell and YSH:
|
| 705 |
|
| 706 | if ! cp foo /tmp {
|
| 707 | echo 'error copying' # any non-zero status
|
| 708 | }
|
| 709 |
|
| 710 | if ! test -d /bin {
|
| 711 | echo 'not a directory'
|
| 712 | }
|
| 713 |
|
| 714 | To be consistent with the idioms below, you can also write them like this:
|
| 715 |
|
| 716 | try {
|
| 717 | cp foo /tmp
|
| 718 | }
|
| 719 | if failed { # shortcut for (_error.code !== 0)
|
| 720 | echo 'error copying'
|
| 721 | }
|
| 722 |
|
| 723 | ### Does a Function Succeed?
|
| 724 |
|
| 725 | When the command is a shell function, you shouldn't use `if myfunc` directly.
|
| 726 | This is because shell has the *Disabled `errexit` Quirk*, which is detected by
|
| 727 | YSH `strict_errexit`.
|
| 728 |
|
| 729 | **No**:
|
| 730 |
|
| 731 | if myfunc; then # errors not checked in body of myfunc
|
| 732 | echo 'success'
|
| 733 | fi
|
| 734 |
|
| 735 | **Yes**. The *`$0` Dispatch Pattern* is a workaround that works in all shells.
|
| 736 |
|
| 737 | if $0 myfunc; then # invoke a new shell
|
| 738 | echo 'success'
|
| 739 | fi
|
| 740 |
|
| 741 | "$@" # Run the function $1 with args $2, $3, ...
|
| 742 |
|
| 743 | **Yes**. The YSH `try` builtin sets the special `_error` variable and returns
|
| 744 | `0`.
|
| 745 |
|
| 746 | try {
|
| 747 | myfunc # doesn't abort
|
| 748 | }
|
| 749 | if failed {
|
| 750 | echo 'success'
|
| 751 | }
|
| 752 |
|
| 753 | ### Does a Pipeline Succeed?
|
| 754 |
|
| 755 | No:
|
| 756 |
|
| 757 | if ps | grep python; then
|
| 758 | echo 'found'
|
| 759 | fi
|
| 760 |
|
| 761 | This is technically correct when `pipefail` is on, but it's impossible for
|
| 762 | YSH `strict_errexit` to distinguish it from `if myfunc | grep python` ahead
|
| 763 | of time (the ["meta" pitfall](error-handling.html#the-meta-pitfall)). If you
|
| 764 | know what you're doing, you can disable `strict_errexit`.
|
| 765 |
|
| 766 | Yes:
|
| 767 |
|
| 768 | try {
|
| 769 | ps | grep python
|
| 770 | }
|
| 771 | if failed {
|
| 772 | echo 'found'
|
| 773 | }
|
| 774 |
|
| 775 | # You can also examine the status of each part of the pipeline
|
| 776 | if (_pipeline_status[0] !== 0) {
|
| 777 | echo 'ps failed'
|
| 778 | }
|
| 779 |
|
| 780 | ### Does a Command With Process Subs Succeed?
|
| 781 |
|
| 782 | Similar to the pipeline example above:
|
| 783 |
|
| 784 | No:
|
| 785 |
|
| 786 | if ! comm <(sort left.txt) <(sort right.txt); then
|
| 787 | echo 'error'
|
| 788 | fi
|
| 789 |
|
| 790 | Yes:
|
| 791 |
|
| 792 | try {
|
| 793 | comm <(sort left.txt) <(sort right.txt)
|
| 794 | }
|
| 795 | if failed {
|
| 796 | echo 'error'
|
| 797 | }
|
| 798 |
|
| 799 | # You can also examine the status of each process sub
|
| 800 | if (_process_sub_status[0] !== 0) {
|
| 801 | echo 'first process sub failed'
|
| 802 | }
|
| 803 |
|
| 804 | (I used `comm` in this example because it doesn't have a true / false / error
|
| 805 | status like `diff`.)
|
| 806 |
|
| 807 | ### Handle Errors in YSH Expressions
|
| 808 |
|
| 809 | try {
|
| 810 | var x = 42 / 0
|
| 811 | echo "result is $[42 / 0]"
|
| 812 | }
|
| 813 | if failed {
|
| 814 | echo 'divide by zero'
|
| 815 | }
|
| 816 |
|
| 817 | ### Test Boolean Statuses, like `grep`, `diff`, `test`
|
| 818 |
|
| 819 | The YSH `boolstatus` builtin distinguishes **error** from **false**.
|
| 820 |
|
| 821 | **No**, this is subtly wrong. `grep` has 3 different return values.
|
| 822 |
|
| 823 | if grep 'class' *.py {
|
| 824 | echo 'found' # status 0 means found
|
| 825 | } else {
|
| 826 | echo 'not found OR ERROR' # any non-zero status
|
| 827 | }
|
| 828 |
|
| 829 | **Yes**. `boolstatus` aborts the program if `egrep` doesn't return 0 or 1.
|
| 830 |
|
| 831 | if boolstatus grep 'class' *.py { # may abort
|
| 832 | echo 'found' # status 0 means found
|
| 833 | } else {
|
| 834 | echo 'not found' # status 1 means not found
|
| 835 | }
|
| 836 |
|
| 837 | More flexible style:
|
| 838 |
|
| 839 | try {
|
| 840 | grep 'class' *.py
|
| 841 | }
|
| 842 | case (_error.code) {
|
| 843 | (0) { echo 'found' }
|
| 844 | (1) { echo 'not found' }
|
| 845 | (else) { echo 'fatal' }
|
| 846 | }
|
| 847 |
|
| 848 | ## Use YSH Expressions, Initializations, and Assignments (var, setvar)
|
| 849 |
|
| 850 | ### Set an Environment Variable Globally
|
| 851 |
|
| 852 | No:
|
| 853 |
|
| 854 | export PYTHONPATH=. # export is disabled in YSH
|
| 855 |
|
| 856 | Yes:
|
| 857 |
|
| 858 | setglobal ENV.PYTHONPATH = '.'
|
| 859 |
|
| 860 | That is, enviroments use the [ENV][] object/namespace, rather than being global
|
| 861 | variables.
|
| 862 |
|
| 863 | [ENV]: ref/chap-special-var.html#ENV
|
| 864 |
|
| 865 | Note: the idiom for setting an env var locally is unchanged:
|
| 866 |
|
| 867 | PYTHONPATH=. myscript.py
|
| 868 |
|
| 869 | ### Initialize and Assign Strings and Integers
|
| 870 |
|
| 871 | No:
|
| 872 |
|
| 873 | local mystr=foo
|
| 874 | mystr='new value'
|
| 875 |
|
| 876 | local myint=42 # still a string in shell
|
| 877 |
|
| 878 | Yes:
|
| 879 |
|
| 880 | var mystr = 'foo'
|
| 881 | setvar mystr = 'new value'
|
| 882 |
|
| 883 | var myint = 42 # a real integer
|
| 884 |
|
| 885 | ### Expressions on Integers
|
| 886 |
|
| 887 | No:
|
| 888 |
|
| 889 | x=$(( 1 + 2*3 ))
|
| 890 | (( x = 1 + 2*3 ))
|
| 891 |
|
| 892 | Yes:
|
| 893 |
|
| 894 | setvar x = 1 + 2*3
|
| 895 |
|
| 896 | ### Mutate Integers
|
| 897 |
|
| 898 | No:
|
| 899 |
|
| 900 | (( i++ )) # interacts poorly with errexit
|
| 901 | i=$(( i+1 ))
|
| 902 |
|
| 903 | Yes:
|
| 904 |
|
| 905 | setvar i += 1 # like Python, with a keyword
|
| 906 |
|
| 907 | ### Initialize and Assign Arrays
|
| 908 |
|
| 909 | Arrays in YSH look like `:| my array |` and `['my', 'array']`.
|
| 910 |
|
| 911 | No:
|
| 912 |
|
| 913 | local -a myarray=(one two three)
|
| 914 | myarray[3]='THREE'
|
| 915 |
|
| 916 | Yes:
|
| 917 |
|
| 918 | var myarray = :| one two three |
|
| 919 | setvar myarray[3] = 'THREE'
|
| 920 |
|
| 921 | var same = ['one', 'two', 'three']
|
| 922 | var typed = [1, 2, true, false, null]
|
| 923 |
|
| 924 |
|
| 925 | ### Initialize and Assign Dicts
|
| 926 |
|
| 927 | Dicts in YSH look like `{key: 'value'}`.
|
| 928 |
|
| 929 | No:
|
| 930 |
|
| 931 | local -A myassoc=(['key']=value ['k2']=v2)
|
| 932 | myassoc['key']=V
|
| 933 |
|
| 934 |
|
| 935 | Yes:
|
| 936 |
|
| 937 | # keys don't need to be quoted
|
| 938 | var myassoc = {key: 'value', k2: 'v2'}
|
| 939 | setvar myassoc['key'] = 'V'
|
| 940 |
|
| 941 | ### Get Values From Arrays and Dicts
|
| 942 |
|
| 943 | No:
|
| 944 |
|
| 945 | local x=${a[i-1]}
|
| 946 | x=${a[i]}
|
| 947 |
|
| 948 | local y=${A['key']}
|
| 949 |
|
| 950 | Yes:
|
| 951 |
|
| 952 | var x = a[i-1]
|
| 953 | setvar x = a[i]
|
| 954 |
|
| 955 | var y = A['key']
|
| 956 |
|
| 957 | ### Conditions and Comparisons
|
| 958 |
|
| 959 | No:
|
| 960 |
|
| 961 | if (( x > 0 )); then
|
| 962 | echo 'positive'
|
| 963 | fi
|
| 964 |
|
| 965 | Yes:
|
| 966 |
|
| 967 | if (x > 0) {
|
| 968 | echo 'positive'
|
| 969 | }
|
| 970 |
|
| 971 | ### Substituting Expressions in Words
|
| 972 |
|
| 973 | No:
|
| 974 |
|
| 975 | echo flag=$((1 + a[i] * 3)) # C-like arithmetic
|
| 976 |
|
| 977 | Yes:
|
| 978 |
|
| 979 | echo flag=$[1 + a[i] * 3] # Arbitrary YSH expressions
|
| 980 |
|
| 981 | # Possible, but a local var might be more readable
|
| 982 | echo flag=$['1' if x else '0']
|
| 983 |
|
| 984 | ## Use [Egg Expressions](eggex.html) instead of Regexes
|
| 985 |
|
| 986 | ### Test for a Match
|
| 987 |
|
| 988 | No:
|
| 989 |
|
| 990 | local pat='[[:digit:]]+'
|
| 991 | if [[ $x =~ $pat ]]; then
|
| 992 | echo 'number'
|
| 993 | fi
|
| 994 |
|
| 995 | Yes:
|
| 996 |
|
| 997 | if (x ~ /digit+/) {
|
| 998 | echo 'number'
|
| 999 | }
|
| 1000 |
|
| 1001 | Or extract the pattern:
|
| 1002 |
|
| 1003 | var pat = / digit+ /
|
| 1004 | if (x ~ pat) {
|
| 1005 | echo 'number'
|
| 1006 | }
|
| 1007 |
|
| 1008 | ### Extract Submatches
|
| 1009 |
|
| 1010 | No:
|
| 1011 |
|
| 1012 | if [[ $x =~ foo-([[:digit:]]+) ]] {
|
| 1013 | echo "${BASH_REMATCH[1]}" # first submatch
|
| 1014 | }
|
| 1015 |
|
| 1016 | Yes:
|
| 1017 |
|
| 1018 | if (x ~ / 'foo-' <capture d+> /) { # <> is capture
|
| 1019 | echo $[_group(1)] # first submatch
|
| 1020 | }
|
| 1021 |
|
| 1022 | ## Glob Matching
|
| 1023 |
|
| 1024 | No:
|
| 1025 |
|
| 1026 | if [[ $x == *.py ]]; then
|
| 1027 | echo 'Python'
|
| 1028 | fi
|
| 1029 |
|
| 1030 | Yes:
|
| 1031 |
|
| 1032 | if (x ~~ '*.py') {
|
| 1033 | echo 'Python'
|
| 1034 | }
|
| 1035 |
|
| 1036 |
|
| 1037 | No:
|
| 1038 |
|
| 1039 | case $x in
|
| 1040 | *.py)
|
| 1041 | echo Python
|
| 1042 | ;;
|
| 1043 | *.sh)
|
| 1044 | echo Shell
|
| 1045 | ;;
|
| 1046 | esac
|
| 1047 |
|
| 1048 | Yes (purely a style preference):
|
| 1049 |
|
| 1050 | case $x { # curly braces
|
| 1051 | (*.py) # balanced parens
|
| 1052 | echo 'Python'
|
| 1053 | ;;
|
| 1054 | (*.sh)
|
| 1055 | echo 'Shell'
|
| 1056 | ;;
|
| 1057 | }
|
| 1058 |
|
| 1059 | ## TODO
|
| 1060 |
|
| 1061 | ### Distinguish Between Variables and Functions
|
| 1062 |
|
| 1063 | - `$RANDOM` vs. `random()`
|
| 1064 | - `LANG=C` vs. `shopt --setattr LANG=C`
|
| 1065 |
|
| 1066 | ## Related Documents
|
| 1067 |
|
| 1068 | - [Shell Language Idioms](shell-idioms.html). This advice applies to shells
|
| 1069 | other than YSH.
|
| 1070 | - [What Breaks When You Upgrade to YSH](upgrade-breakage.html). Shell constructs that YSH
|
| 1071 | users should avoid.
|
| 1072 | - [YSH Fixes Shell's Error Handling (`errexit`)](error-handling.html). YSH fixes the
|
| 1073 | flaky error handling in POSIX shell and bash.
|
| 1074 | - TODO: Go through more of the [Pure Bash
|
| 1075 | Bible](https://github.com/dylanaraps/pure-bash-bible). YSH provides
|
| 1076 | alternatives for such quirky syntax.
|
| 1077 |
|