| Commit message (Collapse) | Author |
|---|
|
Previously we always expanded here-documents in a subshell. This is
contrary to the POSIX specification and how other shells behave. What's
more this slows down many expansions due to the extra fork (however, it
must be said that it is possible for it speed up certain expansions by
running it simultaneously with the command on two CPUs).
This patch move the expansion into the current shell environment.
Test case:
unset a
cat <<- EOF > /dev/null
${a=NOT}
EOF
echo ${a}BAD
Old result:
BAD
New result:
NOTBAD
|
|
The herefd hack goes back more than a decade. it limits the amount of
memory we have to allocate when expanding here-documents by writing the
result out from time to time. However, it's no longer safe because the
stack is used to place intermediate results too and there we certainly
don't want to write them out should we be short on memory.
In any case, with today's computers we can afford to keep the entire
result in memory and write them out at the end.
|
|
On Tue, Oct 30, 2007 at 04:23:35AM +0000, Oleg Verych wrote:
>
> } 8<<""
> ======================
Actually this (the empty delim) only works with dash by accident.
I've tried bash and pdksh and they both terminate on the first
empty line which is what you would expect rather than EOF. The
real Korn shell does something completely different.
I've fixed this in dash to conform to bash/pdksh.
> In [0] it's stated, that delimiter isn't evaluated (expanded), only
> quoiting must be checked. That if() seems to be completely bogus.
OK I agree. The reason it was there is because the parser would
have already replaced the dollar sign by an internal representation.
I've fixed it properly with this patch.
Test case:
cat <<- $a
OK
$a
cat <<- ""
OK
echo OK
Old result:
dash: Syntax error: Illegal eof marker for << redirection
OK
echo OK
New result:
OK
OK
OK
|
|
The change
[PARSER] Recognise here-doc delimiters terminated by EOF
introduced a regerssion whereby lines starting with eofmark but are not equal
to eofmark would be corrupted. This patch fixes it.
Test case:
cat << _ACEOF
_ASBOX
_ACEOF
Old result:
SASBOX
New result:
_ASBOX
|
|
Debian's libc6 as of 2.6.1-6 has working glob(3)/fnmatch(3) support.
This patch adds the options --enable-glob and --enable-fnmatch to
the configure script. By default glob(3) and fnmatch(3) are still
unused. However, on distros where the glibc is known to work you
may enable these options.
|
|
This patch adds the badnum function and uses it to mostly replace the use
of illnum except in miscbltin where the current code turns out to be smaller
because of the twin sh_error calls.
|
|
* NULL as a number argument:
olecom@deen:/mnt/debian/src/dash-0.5.3$ time src/dash tst-01.sh
test: 20: `': bad number
`' eq 0: 2 must be >1, Not A Number
test: 20: `': bad number
`' ne 0: 2 must be >1, Not A Number
test: 20: `': bad number
`' gt 0: 2 must be >1, Not A Number
test: 20: `': bad number
`' ge 0: 2 must be >1, Not A Number
test: 20: `': bad number
`' lt 0: 2 must be >1, Not A Number
test: 20: `': bad number
`' le 0: 2 must be >1, Not A Number
/usr/bin/test: invalid integer `'
`' eq 0: 2 must be >1, Not A Number
/usr/bin/test: invalid integer `'
`' ne 0: 2 must be >1, Not A Number
/usr/bin/test: invalid integer `'
`' gt 0: 2 must be >1, Not A Number
/usr/bin/test: invalid integer `'
`' ge 0: 2 must be >1, Not A Number
/usr/bin/test: invalid integer `'
`' lt 0: 2 must be >1, Not A Number
/usr/bin/test: invalid integer `'
`' le 0: 2 must be >1, Not A Number
#!/usr/bin/printf This not executable script%c\n
test_arithm() {
for aop in eq ne gt ge lt le
do "$1" 0 -$aop "$NOTHING"
echo "\`' $aop 0:" $? " must be >1, Not A Number"
done
}
# opengroup.org/onlinepubs/000095399/utilites/test.html (nothing about long):
test_arithm test
test_arithm /usr/bin/test
# shend
|
|
The function tryexec used the original name instead of the path found through
PATH search. This patch fixes that.
Test case:
trap 'rm -f $TMP' EXIT
TMP=$(tempfile -s nosuchthing)
cat <<- EOF > $TMP
echo OK
EOF
chmod u+x $TMP
cd /
PATH=${TMP%/*} ${TMP##*/}
Old result:
/bin/sh: Can't open filelgY4Fanosuchthing
New result:
OK
|
|
This patch adds a flag argument to setvarint and uses it to set the OPTIND
variable.
|
|
The case where the expansion isn't quoted is the norm.
|
|
Use += instead of straight assignment for token value.
|
|
This patch adds assignment operator support in arithmetic expansions. It
also changes the type used to intmax_t.
|
|
On Wed, Apr 11, 2007 at 01:24:21PM -0700, Micah Cowan wrote:
> Package: dash
> Version: 0.5.3-3
>
> Bug first reported against Ubuntu at
> https://bugs.launchpad.net/ubuntu/+source/dash/+bug/105634
> by Paul Smith
>
> The description and some comments from that bug report follow.
>
> -----
>
> This operation fails on Ubuntu:
>
> $ /bin/sh -c 'if false; then d="${foo/bar}"; fi'
> /bin/sh: Syntax error: Bad substitution
>
> When used with other POSIX shells it succeeds. While semantically the
> variable reference ${foo/bar} is not valid, this is not a syntax error
> according to POSIX, and since the variable assignment expression is
> never invoked (because it's within an "if false") it should not be seen
> as an error.
>
> I ran into this because after restarting my system I could no longer log
> in. It turns out that the problem was (a) I had edited .gnomerc to
> source my .bashrc file so that my environment would be set properly, and
> (b) I had added some new code to my .bashrc WITHIN A CHECK FOR BASH!
> that used bash's ${var/match/sub} feature. Even though this code was
> within a "case $BASH_VERSION; in *[0-9]*) ... esac (so dash would never
> execute it since that variable is not set), it still caused dash to
> throw up.
>
> -----
>
> FYI, some relevant details from POSIX:
>
> Section 2.3, Token Recognition:
>
> 5. If the current character is an unquoted '$' or '`', the shell shall
> identify the start of any candidates for parameter expansion ( Parameter
> Expansion), command substitution ( Command Substitution), or arithmetic
> expansion ( Arithmetic Expansion) from their introductory unquoted
> character sequences: '$' or "${", "$(" or '`', and "$((", respectively.
> The shell shall read sufficient input to determine the end of the unit
> to be expanded (as explained in the cited sections).
>
> Section 2.6.2, Parameter Expansion:
>
> The format for parameter expansion is as follows:
>
> ${expression}
>
> where expression consists of all characters until the matching '}'. Any
> '}' escaped by a backslash or within a quoted string, and characters in
> embedded arithmetic expansions, command substitutions, and variable
> expansions, shall not be examined in determining the matching '}'.
>
> [...]
>
> The parameter name or symbol can be enclosed in braces, which are
> optional except for positional parameters with more than one digit or
> when parameter is followed by a character that could be interpreted as
> part of the name. The matching closing brace shall be determined by
> counting brace levels, skipping over enclosed quoted strings, and
> command substitutions.
>
> ---
>
> In addition to bash I've checked Solaris /bin/sh and ksh and they don't
> report an error.
>
> -----
> Micah Cowan:
>
> The applicable portion of POSIX is in XCU 2.10.1:
>
> "The WORD tokens shall have the word expansion rules applied to them
> immediately before the associated command is executed, not at the time
> the command is parsed."
>
> This seems fairly clear to me.
This patch moves the error detection to expansion time.
Test case:
if false; then
echo ${a!7}
fi
echo OK
Old result:
dash: Syntax error: Bad substitution
New result:
OK
|
|
This patch adds the function atomax10 and uses it in test(1) so that we
support intmax_t comparisons.
|
|
The only user of setvarsafe is getopts. However, we can achieve the same
result by pre-setting the value of shellparam.optind.
|
|
Previously setting OPTIND to 0 would cause subsequent getopts calls to fail.
This patch makes dash reset the getopts parameters the same way as OPTIND=1.
Both behaviours are allowed by POSIX but other common shells do tolerate this
case.
|
|
This patch gets rid of the stack mark tracking hack by allocating a little
bit of stack memory if we're at risk of planting a stack mark which may be
grown later. To do this a new function pushstackmark is added which lets
the user pick a bigger amount to allocate since some users do that anyway
after setting a stack mark.
|
|
The function grabstackblock is identical in semantics to stalloc within its
input constraints.
|
|
Merge flags into subtype.
Do not write subtype out twice.
Add likely flag on ${ vs. $NAME.
Kill unnecessary (and bogus) PEOA check.
|
|
Previously dash treated ${##1} as a length operation. This patch fixes that.
Test case:
set -- a
echo ${##1}OK
Old result:
1OK
New result:
OK
|
|
Add new option to build statical linked dash.
|
|
Previously dash required a <newline> character to be present in order for
a here-document delimiter to be detected. Allowing EOF in the absence of
a <newline> to play the same purpose allows some intuitive scripts to
succeed. POSIX seems to be silence on this so this should be OK.
Test case:
eval 'cat <<- NOT
test
NOT'
echo OK
Old result:
test
NOTOK
New result:
test
OK
|
|
dash-0.5.3 has an issue reading some line lengths [1].
This is reproducable on amd64, but not on other arches for some reason.
$ cat bug.sh
(read line; echo "${line%%=*}") <<EOF
TITLE=xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx+xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
EOF
printf "\
TITLE=xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx+xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
" | (read line; echo "${line%%=*}")
$ bash bug.sh
TITLE
TITLE
$ dash bug.sh
TITLE=xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx+xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
TITLE
Attahced is a patch to fix the issue
Thanks
Roy
[1] http://bugs.gentoo.org/show_bug.cgi?id=180680
|
|
Document that in a for loop 'in word ...' is optional, and if omitted,
'in "$@"' is used.
Lars Wilke noticed this, and reported through
http://bugs.debian.org/387441
|
|
Have the man page explicitely state how the test builtin behaves on
-nt and -ot options if file2 does not exist. The case where file1
does not exist was already documented properly.
This was noticed by Sven Mueller and reported through
http://bugs.debian.org/373611
|
|
Because the parser does not recursively parse parameter expansion with respect
to quotes, we can't accurately determine quote status at parse time. This
patch works around this by moving the quote detection to run-time where we
do interpret it recursively.
Test case:
foo=\\
echo "<${foo#[\\]}>"
Old result:
<\>
New result:
<>
|
|
Previously code was added so that tilde expansion was carried out
parameter expansions within double quotes. This change was made
with reference the behaviour of bash at the time. Bash has since
be fixed so that this behaviour no longer occurs which is in line
with most other POSIX shells.
So this patch removes that behaviour in dash as well.
Test case:
unset a
echo "${a:-~root}"
Old result:
/root
New result:
~root
|
|
Collapsing arithmethc expansion is incorrect when the inner arithmetic
expansion is a part of a parameter expansion.
Test case:
unset a
echo $((3 + ${a:=$((4 + 5))}))
echo $a
Old result:
(4 + 5)
New result:
9
|
|
When an arithmetic expansion terminates and we restore the syntax to the
previous one, we don't need to set dblquote because we never changed upon
entering the arithmetic expansion.
|
|
dqvarnest is only used to determine whether CENDQUOTE should terminate the
double-quote syntax. Since CENDQUOTE can never occur while arinest is set,
we don't need to take arinest into account for dqvarnest.
|
|
If arinest is set then the syntax must be ARISYNTAX. As such CENDQUOTE can
never occur while arinest is set so we don't need to test for it.
|
|
Previously tilde expansion was not carried out for =?#% expansion words.
This is contrary to the POSIX specification.
Test case:
a=~root:~root
echo ${a#~root}
Old result:
/root:/root
New result:
:/root
|
|
Test case:
a=/b/c/*
b=\\
echo ${a%$b*}
Old result:
/b/c/*
New result:
/b/c/
|
|
* Speed up (libc=glibc):
deen:debian/src/dash-0.5.3# echo $(((7853+8631+7529+9777+9161+7552)/6))
8417,8250 # this patch
deen:/mnt/work/debian/src/dash-0.5.3# echo $(((9553+7789+9450+9925+7595+9590)/6))
8983 # short
deen:debian/src/dash-0.5.3# echo $(( (9655+7853+9733+7826+9618+10053)/6 ))
9123 # '[' ']'
deen:debian/src/dash-0.5.3#
deen:debian/src/dash-0.5.3# echo $(((9231+9423+9365+9650+8883+8291)/6))
9140 # unpatched
deen:debian/src/dash-0.5.3#
* Size down:
olecom@deen:/mnt/debian/src/dash-0.5.3$ size src/test.o # this patchset
text data bss dec hex filename
4142 0 16 4158 103e src/test.o
olecom@deen:/mnt/debian/src/dash-0.5.3$ size src/test.o
text data bss dec hex filename
4209 0 16 4225 1081 src/test.o
olecom@deen:/mnt/debian/src/dash-0.5.3$
|
|
Some trailing whitespace was killed or tabified.
|
|
We need to flush at the very end in case we've generated any errors
before that. The flushall call cannot perform a longjmp so it's
safe there.
|
|
On Thu, Jun 07, 2007 at 12:20:27PM +0200, Denis Vlasenko wrote:
>
> When I start dash under Midnight Commander and then type 'exit', dash
> exits all right, but then MC is sent to background. It happens because
> dash does not restore current process group on exit.
>
> Attached patch fixes this. It also fixes another bug: setjobctl(0)
> must ignore tcsetpgrp errors, because there are cases when tty is
> destroyed under dash.
>
> Patch is run-tested.
I've fixed this slightly differently so that we don't need the
xtcsetpgrp change.
|
|
|
|
Add likely flag on REDIR_PUSH.
Remove sv check on REDIR_SAVEFD2 (it implies REDIR_PUSH).
Optimise NTOFD/NFROMFD same fd check.
|
|
As it stands if a redirection occurs on a file descriptor that was
previously closed, it won't be closed after that redirection goes
out of effect. This is because we don't keep track of closed file
descriptors properly as we do for open ones.
This patch fixes this by introducing two new states, CLOSED and
REALLY_CLOSED. The first represents an initially closed descriptor
which is now open while the second one represents an initally closed
descriptor which is now closed.
This patch is based on work by Rainer Weikusat.
|
|
Now that we're marking file descriptors as CLOEXEC in savefd, we no longer
need to close them on exec or in setinputfd.
|
|
There are two kinds of users to copyfd, those that want to copy an fd to
an exact value and those that want to move an fd to a value >= 10. The
former can simply use dup2 directly while the latter share a lot of common
code that now constitutes savefd.
|
|
Since we know that dup2 must succeed here we can call it directly.
|
|
No caller of copyfd need to ignore EMFILE so we can remove the special
case and just let it call sh_error on any error.
|
|
|
|
The value for CTLBACKQ | CTLQUOTE is now -123, not 133.
|
|
On Fri, Feb 16, 2007 at 04:24:55PM -0800, Dan Nicholson wrote:
> I was having a problem building dash where `:' wasn't being recognized
> as a builtin command. I won't bore you with the details of the
> debugging, but it came down to the fact that the generated struct in
> builtins.c wasn't being sorted correctly. The ":" name was coming
> before the "." name and then was never able to be found by bsearch().
>
> The issue turned out to be that I was building with a UTF-8 charset,
> causing the `sort' to come out differently from the mkbuiltins script.
> When I finally set LANG=C, it came out correct. Here's an example of
> the difference.
>
> $ LANG=C sort << "EOF"
> :
> .
> EOF
> .
> :
> $ LANG=en_US.ISO-8859-1 sort << "EOF"
> :
> .
> EOF
> :
> .
>
> My system has glibc-2.3.6, but I also tested it on an RHEL3 server at
> work. My solution was just to add LC_ALL=C to the mkbuiltins script.
> Maybe it's overkill to use LC_ALL. Patch against HEAD attached.
I've changed this patch so that we just set LC_COLLATE for the sort
command.
|
|
On Sat, Oct 21, 2006 at 02:19:18PM +0000, Gerrit Pape wrote:
> Hi Herbert, please see
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=387458
>
> On Thu, Sep 14, 2006 at 03:50:02PM +0200, Julien Danjou wrote:
> > I just found this bug which is easily reproductible:
> >
> > % dash -c 'command -v'
> > zsh: segmentation fault dash -c 'command -v'
Since bash's behaviour is minimalist here, I've decided to adopt its
behaviour here as well which is to return success silently.
|
|
Check getgroups() and fwrite() return code, required to build with
-D_FORTIFY_SOURCE=2.
|
|
The return value of ccmatch was being treated as 0 or 1 but
it's actually zero or non-zero. This broke inverted character
class matching.
Reported by Alexander Skwar.
|