1 |
|
2 |
#### >&
|
3 |
echo hi 1>&2
|
4 |
## stderr: hi
|
5 |
|
6 |
#### <&
|
7 |
# Is there a simpler test case for this?
|
8 |
echo foo > $TMP/lessamp.txt
|
9 |
exec 6< $TMP/lessamp.txt
|
10 |
read line <&6
|
11 |
echo "[$line]"
|
12 |
## stdout: [foo]
|
13 |
|
14 |
#### Leading redirect
|
15 |
echo hello >$TMP/hello.txt # temporary fix
|
16 |
<$TMP/hello.txt cat
|
17 |
## stdout: hello
|
18 |
|
19 |
#### Nonexistent file
|
20 |
cat <$TMP/nonexistent.txt
|
21 |
echo status=$?
|
22 |
## stdout: status=1
|
23 |
## OK dash stdout: status=2
|
24 |
|
25 |
#### Redirect in command sub
|
26 |
FOO=$(echo foo 1>&2)
|
27 |
echo $FOO
|
28 |
## stdout:
|
29 |
## stderr: foo
|
30 |
|
31 |
#### Redirect in assignment
|
32 |
# dash captures stderr to a file here, which seems correct. Bash doesn't and
|
33 |
# just lets it go to actual stderr.
|
34 |
# For now we agree with dash/mksh, since it involves fewer special cases in the
|
35 |
# code.
|
36 |
|
37 |
FOO=$(echo foo 1>&2) 2>$TMP/no-command.txt
|
38 |
echo FILE=
|
39 |
cat $TMP/no-command.txt
|
40 |
echo "FOO=$FOO"
|
41 |
## STDOUT:
|
42 |
FILE=
|
43 |
foo
|
44 |
FOO=
|
45 |
## END
|
46 |
## BUG bash STDOUT:
|
47 |
FILE=
|
48 |
FOO=
|
49 |
## END
|
50 |
|
51 |
#### Redirect in function body.
|
52 |
fun() { echo hi; } 1>&2
|
53 |
fun
|
54 |
## STDOUT:
|
55 |
## END
|
56 |
## STDERR:
|
57 |
hi
|
58 |
## END
|
59 |
|
60 |
#### Redirect in function body is evaluated multiple times
|
61 |
i=0
|
62 |
fun() { echo "file $i"; } 1> "$TMP/file$((i++))"
|
63 |
fun
|
64 |
fun
|
65 |
echo i=$i
|
66 |
echo __
|
67 |
cat $TMP/file0
|
68 |
echo __
|
69 |
cat $TMP/file1
|
70 |
## STDOUT:
|
71 |
i=2
|
72 |
__
|
73 |
file 1
|
74 |
__
|
75 |
file 2
|
76 |
## END
|
77 |
## N-I dash stdout-json: ""
|
78 |
## N-I dash status: 2
|
79 |
|
80 |
#### Redirect in function body AND function call
|
81 |
fun() { echo hi; } 1>&2
|
82 |
fun 2>&1
|
83 |
## stdout-json: "hi\n"
|
84 |
## stderr-json: ""
|
85 |
|
86 |
#### Descriptor redirect with spaces
|
87 |
# Hm this seems like a failure of lookahead! The second thing should look to a
|
88 |
# file-like thing.
|
89 |
# I think this is a posix issue.
|
90 |
# tag: posix-issue
|
91 |
echo one 1>&2
|
92 |
echo two 1 >&2
|
93 |
echo three 1>& 2
|
94 |
## stderr-json: "one\ntwo 1\nthree\n"
|
95 |
|
96 |
#### Filename redirect with spaces
|
97 |
# This time 1 *is* a descriptor, not a word. If you add a space between 1 and
|
98 |
# >, it doesn't work.
|
99 |
echo two 1> $TMP/file-redir1.txt
|
100 |
cat $TMP/file-redir1.txt
|
101 |
## stdout: two
|
102 |
|
103 |
#### Quoted filename redirect with spaces
|
104 |
# POSIX makes node of this
|
105 |
echo two \1 > $TMP/file-redir2.txt
|
106 |
cat $TMP/file-redir2.txt
|
107 |
## stdout: two 1
|
108 |
|
109 |
#### Descriptor redirect with filename
|
110 |
# bash/mksh treat this like a filename, not a descriptor.
|
111 |
# dash aborts.
|
112 |
echo one 1>&$TMP/nonexistent-filename__
|
113 |
echo "status=$?"
|
114 |
## stdout: status=1
|
115 |
## BUG bash stdout: status=0
|
116 |
## OK dash stdout-json: ""
|
117 |
## OK dash status: 2
|
118 |
|
119 |
#### redirect for loop
|
120 |
for i in $(seq 3)
|
121 |
do
|
122 |
echo $i
|
123 |
done > $TMP/redirect-for-loop.txt
|
124 |
cat $TMP/redirect-for-loop.txt
|
125 |
## stdout-json: "1\n2\n3\n"
|
126 |
|
127 |
#### redirect subshell
|
128 |
( echo foo ) 1>&2
|
129 |
## stderr: foo
|
130 |
## stdout-json: ""
|
131 |
|
132 |
#### Prefix redirect for loop -- not allowed
|
133 |
>$TMP/redirect2.txt for i in $(seq 3)
|
134 |
do
|
135 |
echo $i
|
136 |
done
|
137 |
cat $TMP/redirect2.txt
|
138 |
## status: 2
|
139 |
## OK mksh status: 1
|
140 |
|
141 |
#### Brace group redirect
|
142 |
# Suffix works, but prefix does NOT work.
|
143 |
# That comes from '| compound_command redirect_list' in the grammar!
|
144 |
{ echo block-redirect; } > $TMP/br.txt
|
145 |
cat $TMP/br.txt | wc -c
|
146 |
## stdout: 15
|
147 |
|
148 |
#### Redirect echo to stderr, and then redirect all of stdout somewhere.
|
149 |
{ echo foo 1>&2; echo 012345789; } > $TMP/block-stdout.txt
|
150 |
cat $TMP/block-stdout.txt | wc -c
|
151 |
## stderr: foo
|
152 |
## stdout: 10
|
153 |
|
154 |
#### Redirect in the middle of two assignments
|
155 |
FOO=foo >$TMP/out.txt BAR=bar printenv.py FOO BAR
|
156 |
tac $TMP/out.txt
|
157 |
## stdout-json: "bar\nfoo\n"
|
158 |
|
159 |
#### Redirect in the middle of a command
|
160 |
f=$TMP/out
|
161 |
echo -n 1 2 '3 ' > $f
|
162 |
echo -n 4 5 >> $f '6 '
|
163 |
echo -n 7 >> $f 8 '9 '
|
164 |
echo -n >> $f 1 2 '3 '
|
165 |
echo >> $f -n 4 5 '6 '
|
166 |
cat $f
|
167 |
## stdout-json: "1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 "
|
168 |
|
169 |
#### Named file descriptor
|
170 |
exec {myfd}> $TMP/named-fd.txt
|
171 |
echo named-fd-contents >& $myfd
|
172 |
cat $TMP/named-fd.txt
|
173 |
## stdout: named-fd-contents
|
174 |
## status: 0
|
175 |
## N-I dash/mksh stdout-json: ""
|
176 |
## N-I dash/mksh status: 127
|
177 |
|
178 |
#### Double digit fd (20> file)
|
179 |
exec 20> "$TMP/double-digit-fd.txt"
|
180 |
echo hello20 >&20
|
181 |
cat "$TMP/double-digit-fd.txt"
|
182 |
## stdout: hello20
|
183 |
## BUG dash stdout-json: ""
|
184 |
## BUG dash status: 127
|
185 |
|
186 |
#### : 9> fdleak (OSH regression)
|
187 |
true 9> "$TMP/fd.txt"
|
188 |
( echo world >&9 )
|
189 |
cat "$TMP/fd.txt"
|
190 |
## stdout-json: ""
|
191 |
|
192 |
#### : 3>&3 (OSH regression)
|
193 |
|
194 |
# mksh started being flaky on the continuous build and during release. We
|
195 |
# don't care! Related to issue #330.
|
196 |
case $SH in (mksh) exit ;; esac
|
197 |
|
198 |
: 3>&3
|
199 |
echo hello
|
200 |
## stdout: hello
|
201 |
## BUG mksh stdout-json: ""
|
202 |
## BUG mksh status: 0
|
203 |
|
204 |
#### : 3>&3-
|
205 |
: 3>&3-
|
206 |
echo hello
|
207 |
## stdout: hello
|
208 |
## N-I dash/mksh stdout-json: ""
|
209 |
## N-I mksh status: 1
|
210 |
## N-I dash status: 2
|
211 |
|
212 |
#### 3>&- << EOF (OSH regression: fail to restore fds)
|
213 |
exec 3> "$TMP/fd.txt"
|
214 |
echo hello 3>&- << EOF
|
215 |
EOF
|
216 |
echo world >&3
|
217 |
exec 3>&- # close
|
218 |
cat "$TMP/fd.txt"
|
219 |
## STDOUT:
|
220 |
hello
|
221 |
world
|
222 |
## END
|
223 |
|
224 |
#### Open file on descriptor 3 and write to it many times
|
225 |
|
226 |
# different than case below because 3 is the likely first FD of open()
|
227 |
|
228 |
exec 3> "$TMP/fd3.txt"
|
229 |
echo hello >&3
|
230 |
echo world >&3
|
231 |
exec 3>&- # close
|
232 |
cat "$TMP/fd3.txt"
|
233 |
## STDOUT:
|
234 |
hello
|
235 |
world
|
236 |
## END
|
237 |
|
238 |
#### Open file on descriptor 4 and write to it many times
|
239 |
|
240 |
# different than the case above because because 4 isn't the likely first FD
|
241 |
|
242 |
exec 4> "$TMP/fd4.txt"
|
243 |
echo hello >&4
|
244 |
echo world >&4
|
245 |
exec 4>&- # close
|
246 |
cat "$TMP/fd4.txt"
|
247 |
## STDOUT:
|
248 |
hello
|
249 |
world
|
250 |
## END
|
251 |
|
252 |
#### Redirect function stdout
|
253 |
f() { echo one; echo two; }
|
254 |
f > $TMP/redirect-func.txt
|
255 |
cat $TMP/redirect-func.txt
|
256 |
## stdout-json: "one\ntwo\n"
|
257 |
|
258 |
#### Nested function stdout redirect
|
259 |
# Shows that a stack is necessary.
|
260 |
inner() {
|
261 |
echo i1
|
262 |
echo i2
|
263 |
}
|
264 |
outer() {
|
265 |
echo o1
|
266 |
inner > $TMP/inner.txt
|
267 |
echo o2
|
268 |
}
|
269 |
outer > $TMP/outer.txt
|
270 |
cat $TMP/inner.txt
|
271 |
echo --
|
272 |
cat $TMP/outer.txt
|
273 |
## stdout-json: "i1\ni2\n--\no1\no2\n"
|
274 |
|
275 |
#### Redirect to empty string
|
276 |
f=''
|
277 |
echo s > "$f"
|
278 |
echo "result=$?"
|
279 |
set -o errexit
|
280 |
echo s > "$f"
|
281 |
echo DONE
|
282 |
## stdout: result=1
|
283 |
## status: 1
|
284 |
## OK dash stdout: result=2
|
285 |
## OK dash status: 2
|
286 |
|
287 |
#### Redirect to file descriptor that's not open
|
288 |
# Notes:
|
289 |
# - 7/2021: descriptor 7 seems to work on all CI systems. The process state
|
290 |
# isn't clean, but we could probably close it in OSH?
|
291 |
# - dash doesn't allow file descriptors greater than 9. (This is a good
|
292 |
# thing, because the bash chapter in AOSA book mentions that juggling user
|
293 |
# vs. system file descriptors is a huge pain.)
|
294 |
# - But somehow running in parallel under spec-runner.sh changes whether
|
295 |
# descriptor 3 is open. e.g. 'echo hi 1>&3'. Possibly because of
|
296 |
# /usr/bin/time. The _tmp/spec/*.task.txt file gets corrupted!
|
297 |
# - Oh this is because I use time --output-file. That opens descriptor 3. And
|
298 |
# then time forks the shell script. The file descriptor table is inherited.
|
299 |
# - You actually have to set the file descriptor to something. What do
|
300 |
# configure and debootstrap too?
|
301 |
|
302 |
opened=$(ls /proc/$$/fd)
|
303 |
if echo "$opened" | egrep '^7$'; then
|
304 |
echo "FD 7 shouldn't be open"
|
305 |
echo "OPENED:"
|
306 |
echo "$opened"
|
307 |
fi
|
308 |
|
309 |
echo hi 1>&7
|
310 |
## stdout-json: ""
|
311 |
## status: 1
|
312 |
## OK dash status: 2
|
313 |
|
314 |
#### Open descriptor with exec
|
315 |
# What is the point of this? ./configure scripts and debootstrap use it.
|
316 |
exec 3>&1
|
317 |
echo hi 1>&3
|
318 |
## stdout: hi
|
319 |
## status: 0
|
320 |
|
321 |
#### Open multiple descriptors with exec
|
322 |
# What is the point of this? ./configure scripts and debootstrap use it.
|
323 |
exec 3>&1
|
324 |
exec 4>&1
|
325 |
echo three 1>&3
|
326 |
echo four 1>&4
|
327 |
## stdout-json: "three\nfour\n"
|
328 |
## status: 0
|
329 |
|
330 |
#### >| to clobber
|
331 |
echo XX >| $TMP/c.txt
|
332 |
|
333 |
set -o noclobber
|
334 |
|
335 |
echo YY > $TMP/c.txt # not clobber
|
336 |
echo status=$?
|
337 |
|
338 |
cat $TMP/c.txt
|
339 |
echo ZZ >| $TMP/c.txt
|
340 |
|
341 |
cat $TMP/c.txt
|
342 |
## STDOUT:
|
343 |
status=1
|
344 |
XX
|
345 |
ZZ
|
346 |
## END
|
347 |
## OK dash STDOUT:
|
348 |
status=2
|
349 |
XX
|
350 |
ZZ
|
351 |
## END
|
352 |
|
353 |
#### &> redirects stdout and stderr
|
354 |
tmp="$(basename $SH)-$$.txt" # unique name for shell and test case
|
355 |
#echo $tmp
|
356 |
|
357 |
stdout_stderr.py &> $tmp
|
358 |
|
359 |
# order is indeterminate
|
360 |
grep STDOUT $tmp
|
361 |
grep STDERR $tmp
|
362 |
|
363 |
## STDOUT:
|
364 |
STDOUT
|
365 |
STDERR
|
366 |
## END
|
367 |
## N-I dash stdout: STDOUT
|
368 |
## N-I dash stderr: STDERR
|
369 |
## N-I dash status: 1
|
370 |
|
371 |
#### >&word redirects stdout and stderr when word is not a number or -
|
372 |
|
373 |
# dash, mksh don't implement this bash behaviour.
|
374 |
case $SH in (dash|mksh) exit 1 ;; esac
|
375 |
|
376 |
tmp="$(basename $SH)-$$.txt" # unique name for shell and test case
|
377 |
|
378 |
stdout_stderr.py >&$tmp
|
379 |
|
380 |
# order is indeterminate
|
381 |
grep STDOUT $tmp
|
382 |
grep STDERR $tmp
|
383 |
|
384 |
## STDOUT:
|
385 |
STDOUT
|
386 |
STDERR
|
387 |
## END
|
388 |
## N-I dash/mksh status: 1
|
389 |
## N-I dash/mksh stdout-json: ""
|
390 |
|
391 |
#### 1>&- to close file descriptor
|
392 |
exec 5> "$TMP/f.txt"
|
393 |
echo hello >&5
|
394 |
exec 5>&-
|
395 |
echo world >&5
|
396 |
cat "$TMP/f.txt"
|
397 |
## stdout-json: "hello\n"
|
398 |
|
399 |
#### 1>&2- to move file descriptor
|
400 |
exec 5> "$TMP/f.txt"
|
401 |
echo hello5 >&5
|
402 |
exec 6>&5-
|
403 |
echo world5 >&5
|
404 |
echo world6 >&6
|
405 |
exec 6>&-
|
406 |
cat "$TMP/f.txt"
|
407 |
## stdout-json: "hello5\nworld6\n"
|
408 |
## N-I dash status: 2
|
409 |
## N-I dash stdout-json: ""
|
410 |
## N-I mksh status: 1
|
411 |
## N-I mksh stdout-json: ""
|
412 |
|
413 |
#### 1>&2- (Bash bug: fail to restore closed fd)
|
414 |
|
415 |
# 7/2021: descriptor 8 is open on Github Actions, so use descriptor 6 instead
|
416 |
|
417 |
# Fix for CI systems where process state isn't clean: Close descriptors 6 and 7.
|
418 |
exec 6>&- 7>&-
|
419 |
|
420 |
opened=$(ls /proc/$$/fd)
|
421 |
if echo "$opened" | egrep '^7$'; then
|
422 |
echo "FD 7 shouldn't be open"
|
423 |
echo "OPENED:"
|
424 |
echo "$opened"
|
425 |
fi
|
426 |
if echo "$opened" | egrep '^6$'; then
|
427 |
echo "FD 6 shouldn't be open"
|
428 |
echo "OPENED:"
|
429 |
echo "$opened"
|
430 |
fi
|
431 |
|
432 |
exec 7> "$TMP/f.txt"
|
433 |
: 6>&7 7>&-
|
434 |
echo hello >&7
|
435 |
: 6>&7-
|
436 |
echo world >&7
|
437 |
exec 7>&-
|
438 |
cat "$TMP/f.txt"
|
439 |
## status: 2
|
440 |
## stdout-json: ""
|
441 |
## OK mksh status: 1
|
442 |
## BUG bash status: 0
|
443 |
## BUG bash stdout: hello
|
444 |
|
445 |
#### <> for read/write
|
446 |
echo first >$TMP/rw.txt
|
447 |
exec 8<>$TMP/rw.txt
|
448 |
read line <&8
|
449 |
echo line=$line
|
450 |
echo second 1>&8
|
451 |
echo CONTENTS
|
452 |
cat $TMP/rw.txt
|
453 |
## stdout-json: "line=first\nCONTENTS\nfirst\nsecond\n"
|
454 |
|
455 |
#### <> for read/write named pipes
|
456 |
rm -f "$TMP/f.pipe"
|
457 |
mkfifo "$TMP/f.pipe"
|
458 |
exec 8<> "$TMP/f.pipe"
|
459 |
echo first >&8
|
460 |
echo second >&8
|
461 |
read line1 <&8
|
462 |
read line2 <&8
|
463 |
exec 8<&-
|
464 |
echo line1=$line1 line2=$line2
|
465 |
## stdout: line1=first line2=second
|
466 |
|
467 |
#### &>> appends stdout and stderr
|
468 |
|
469 |
# Fix for flaky tests: dash behaves non-deterministically under load! It
|
470 |
# doesn't implement the behavior anyway so I don't care why.
|
471 |
case $SH in
|
472 |
*dash)
|
473 |
exit 1
|
474 |
;;
|
475 |
esac
|
476 |
|
477 |
echo "ok" > $TMP/f.txt
|
478 |
stdout_stderr.py &>> $TMP/f.txt
|
479 |
grep ok $TMP/f.txt >/dev/null && echo 'ok'
|
480 |
grep STDOUT $TMP/f.txt >/dev/null && echo 'ok'
|
481 |
grep STDERR $TMP/f.txt >/dev/null && echo 'ok'
|
482 |
## STDOUT:
|
483 |
ok
|
484 |
ok
|
485 |
ok
|
486 |
## END
|
487 |
## N-I dash stdout-json: ""
|
488 |
## N-I dash status: 1
|
489 |
|
490 |
#### exec redirect then various builtins
|
491 |
exec 5>$TMP/log.txt
|
492 |
echo hi >&5
|
493 |
set -o >&5
|
494 |
echo done
|
495 |
## STDOUT:
|
496 |
done
|
497 |
## END
|
498 |
|
499 |
#### >$file touches a file
|
500 |
rm -f myfile
|
501 |
test -f myfile
|
502 |
echo status=$?
|
503 |
>myfile
|
504 |
test -f myfile
|
505 |
echo status=$?
|
506 |
## STDOUT:
|
507 |
status=1
|
508 |
status=0
|
509 |
## END
|
510 |
# regression for OSH
|
511 |
## stderr-json: ""
|
512 |
|
513 |
#### $(< $file) yields the contents of the file
|
514 |
|
515 |
echo FOO > myfile
|
516 |
foo=$(< myfile)
|
517 |
echo $foo
|
518 |
## STDOUT:
|
519 |
FOO
|
520 |
## END
|
521 |
## N-I dash/ash/yash stdout-json: "\n"
|
522 |
|
523 |
#### $(< file) with more statements
|
524 |
|
525 |
# note that it doesn't do this without a command sub!
|
526 |
# It's apparently a special case in bash, mksh, and zsh?
|
527 |
foo=$(echo begin; < myfile)
|
528 |
echo $foo
|
529 |
echo ---
|
530 |
|
531 |
foo=$(< myfile; echo end)
|
532 |
echo $foo
|
533 |
echo ---
|
534 |
|
535 |
foo=$(< myfile; <myfile)
|
536 |
echo $foo
|
537 |
echo ---
|
538 |
|
539 |
## STDOUT:
|
540 |
begin
|
541 |
---
|
542 |
end
|
543 |
---
|
544 |
|
545 |
---
|
546 |
## END
|
547 |
# weird, zsh behaves differently
|
548 |
## OK zsh STDOUT:
|
549 |
begin
|
550 |
FOO
|
551 |
---
|
552 |
FOO
|
553 |
end
|
554 |
---
|
555 |
FOO
|
556 |
FOO
|
557 |
---
|
558 |
## END
|
559 |
|
560 |
|
561 |
#### < file in pipeline and subshell doesn't work
|
562 |
echo FOO > file2
|
563 |
|
564 |
# This only happens in command subs, which is weird
|
565 |
< file2 | tr A-Z a-z
|
566 |
( < file2 )
|
567 |
echo end
|
568 |
## STDOUT:
|
569 |
end
|
570 |
## END
|
571 |
|
572 |
#### 2>&1 with no command
|
573 |
( exit 42 ) # status is reset after this
|
574 |
echo status=$?
|
575 |
2>&1
|
576 |
echo status=$?
|
577 |
## STDOUT:
|
578 |
status=42
|
579 |
status=0
|
580 |
## END
|
581 |
## stderr-json: ""
|
582 |
|
583 |
#### 2&>1 (is it a redirect or is it like a&>1)
|
584 |
2&>1
|
585 |
echo status=$?
|
586 |
## STDOUT:
|
587 |
status=127
|
588 |
## END
|
589 |
## OK mksh/dash STDOUT:
|
590 |
status=0
|
591 |
## END
|
592 |
|
593 |
#### can't mention big file descriptor
|
594 |
echo hi 9>&1
|
595 |
# trivia: 23 is the max descriptor for mksh
|
596 |
#echo hi 24>&1
|
597 |
echo hi 99>&1
|
598 |
echo hi 100>&1
|
599 |
## OK osh STDOUT:
|
600 |
hi
|
601 |
hi
|
602 |
hi 100
|
603 |
## END
|
604 |
## STDOUT:
|
605 |
hi
|
606 |
hi 99
|
607 |
hi 100
|
608 |
## END
|
609 |
## BUG bash STDOUT:
|
610 |
hi
|
611 |
hi
|
612 |
hi
|
613 |
## END
|
614 |
|
615 |
#### : >/dev/null 2> / (OSH regression: fail to pop fd frame)
|
616 |
# oil 0.8.pre4 fails to restore fds after redirection failure. In the
|
617 |
# following case, the fd frame remains after the redirection failure
|
618 |
# "2> /" so that the effect of redirection ">/dev/null" remains after
|
619 |
# the completion of the command.
|
620 |
: >/dev/null 2> /
|
621 |
echo hello
|
622 |
## stdout: hello
|
623 |
## OK dash stdout-json: ""
|
624 |
## OK dash status: 2
|
625 |
## OK mksh stdout-json: ""
|
626 |
## OK mksh status: 1
|
627 |
# dash/mksh terminates the execution of script on the redirection.
|
628 |
|
629 |
#### echo foo >&100 (OSH regression: does not fail with invalid fd 100)
|
630 |
# oil 0.8.pre4 does not fail with non-existent fd 100.
|
631 |
fd=100
|
632 |
echo foo >&$fd
|
633 |
## stdout-json: ""
|
634 |
## status: 1
|
635 |
## OK dash status: 2
|
636 |
|
637 |
#### echo foo >&N where N is first unused fd
|
638 |
# 1. prepare default fd for internal uses
|
639 |
minfd=10
|
640 |
case ${SH##*/} in
|
641 |
(mksh) minfd=24 ;;
|
642 |
(osh) minfd=100 ;;
|
643 |
esac
|
644 |
|
645 |
# 2. prepare first unused fd
|
646 |
fd=$minfd
|
647 |
is-fd-open() { : >&$1; }
|
648 |
while is-fd-open "$fd"; do
|
649 |
: $((fd+=1))
|
650 |
|
651 |
# prevent infinite loop for broken oils-for-unix
|
652 |
if test $fd -gt 1000; then
|
653 |
break
|
654 |
fi
|
655 |
done
|
656 |
|
657 |
# 3. test
|
658 |
echo foo >&$fd
|
659 |
## stdout-json: ""
|
660 |
## status: 1
|
661 |
## OK dash status: 2
|
662 |
|
663 |
#### exec {fd}>&- (OSH regression: fails to close fd)
|
664 |
# mksh, dash do not implement {fd} redirections.
|
665 |
case $SH in (mksh|dash) exit 1 ;; esac
|
666 |
# oil 0.8.pre4 fails to close fd by {fd}&-.
|
667 |
exec {fd}>file1
|
668 |
echo foo >&$fd
|
669 |
exec {fd}>&-
|
670 |
echo bar >&$fd
|
671 |
cat file1
|
672 |
## stdout: foo
|
673 |
## N-I mksh/dash stdout-json: ""
|
674 |
## N-I mksh/dash status: 1
|