mirror/ zsh
1
0
Fork 0
mirror of git://git.code.sf.net/p/zsh/code synced 2024-05-12 10:36:17 +02:00
Code

52558: word splitting differences of nofork; update ToC; formatting fixes

This commit is contained in:
Bart Schaefer 2024-02-18 14:25:20 -08:00
parent 336249e7ea
commit 4a86a54d2b
2 changed files with 41 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
ChangeLog
View File

@ -1,3 +1,8 @@
2024-02-18 Bart Schaefer <schaefer@toltec-ubuntu>
* 52558: Etc/FAQ.yo: make note of word splitting differences
with nofork substitutions; update ToC; minor formatting fixes
2024-02-18 Stephane Chazelas <stephane@chazelas.org>
* 45837: Src/exec.c, Src/init.c, Src/options.c: fix issue whereby

60
Etc/FAQ.yo
View File

@ -101,6 +101,10 @@ Chapter 2: How does zsh differ from...?
2.6. Shouldn't zsh be more/less like ksh/(t)csh?
2.7. What is zsh's support for Unicode/UTF-8?
2.8. Why does my bash script report an error when I run it under zsh?
2.9. What is a `namespace' anyway?
2.10. What about named references?
2.11. What is zsh's support for non-forking command substitution?
2.12. Comparisons of forking and non-forking command substitution
Chapter 3: How to get various things to work
3.1. Why does `$var' where `var="foo bar"' not do what I expect?
@ -1029,27 +1033,32 @@ label(211)
This is for cases where you'd write mytt($(command)) but you don't want
the overhead or other issues associated with forking a subshell.
There are 3 variations:
enumeration(
myeit() Borrowed from mksh
itemization(
eit() Borrowed from mksh
verb(
${| code }
)
Runs code in the current shell context and then substitutes mytt(${REPLY}).
The result is not split into words unless the tt(SH_WORD_SPLIT) option
is set, for example by mytt(${=${| code }}).
myeit() An extension to #1
eit() An extension to #1
verb(
${|var| code }
)
Runs code in the current shell and then substitutes mytt(${var}).
Runs code in the current shell and then substitutes mytt(${var}). If
mytt(${var}) names an array, the result is an array of those elements,
but no further splitting is done without tt(SH_WORD_SPLIT).
myeit() The traditional ksh form, except that the closing mytt(;)
eit() The traditional ksh form, except that the closing mytt(;)
may usually be omitted:
verb(
${ code }
)
Runs code in the current shell and substitutes its standard output.
(this is done with a temporary file ala mytt($(<=( code ))) but
without the fork implied by mytt(=(...))).
without the fork implied by mytt(=(...))). The result is not split
into words without tt(SH_WORD_SPLIT).
)
In all three forms mytt(code) behaves myem(similarly) to an anonymous
@ -1079,31 +1088,34 @@ sect(Comparisons of forking and non-forking command substitution)
when substituting, whereas mytt(${ command }) and its variants do not.
The latter is consistent with mytt(${|...}) from mksh but differs from
bash and ksh, so in emulation modes, newlines are stripped from command
output (not from mytt(REPLY) assignments).
output (not from tt(REPLY) assignments).
When not enclosed in double quotes, the expansion of mytt($(command)) is
split on tt(IFS) into an array of words. In contrast, and unlike both
bash and ksh, unquoted non-forking substitutions behave like parameter
expansions with respect to the tt(SH_WORD_SPLIT) option.
When mytt(command) is myem(not) a builtin, mytt(${ command }) does fork, and
typically forks the same number of times as mytt($(command)), because in
the latter case zsh usually optimizes the final fork into an exec.
Redirecting input from files has subtle differences:
mytt($(<file)) is optimized to read from mytt(file) without forking, but
per above it removes trailing newlines.
mytt(${<file}) is a substitution error.
mytt(${ <file }) copies mytt(file) using the mytt(NULLCMD) programs, then
reads and substitutes the contents of the copy. Also, this fails if the
mytt(CSH_NULLCMD) or mytt(SH_NULLCMD) options are in effect, so it does
not work in emulation modes.
mytt(${|<file}) copies mytt(file) to the standard output using mytt(NULLCMD)
but substitutes nothing because there is no assignment to mytt(REPLY). It
fails in emulation modes.
itemization(
it() mytt($(<file)) is optimized to read from mytt(file) without forking,
but per above it removes trailing newlines.
it() mytt(${<file}) is a substitution error.
it() mytt(${ <file }) copies mytt(file) using the mytt(NULLCMD) programs,
then reads and substitutes the contents of the copy. Also, this
fails if the tt(CSH_NULLCMD) or tt(SH_NULLCMD) options are in effect,
so it does not work in emulation modes.
it() mytt(${|<file}) copies mytt(file) to the standard output using
tt(NULLCMD) but substitutes nothing because there is no assignment
to tt(REPLY). It also fails in emulation modes.
)
mytt(${|IFS= read -rd '' <file}) is therefore the best solution for files
that do not contain nul bytes, because it copies the file directly into
the local mytt(REPLY) and then substitutes that.
the local mytt(REPLY) and then substitutes that. For very large files,
refer to mytt(Functions/Misc/zslurp).
chapter(How to get various things to work)
@ -1979,7 +1991,7 @@ sect(Why is my output duplicated with `tt(foo 2>&1 >foo.out | bar)'?)
to both files when the redirector appears twice. What's going on in the
first example is exactly the same, however the second redirector is
disguised as a pipe. So if you want to turn this effect off, you need
to unset the option mytt(MULTIOS), or alternatively write the following:
to unset the option tt(MULTIOS), or alternatively write the following:
verb(
% { print output; print error >&2 } 2>&1 >&- >foo.out | sed 's/error/erratic/'
erratic