JehanneOS/mksh
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

fix exit-eval-1 (Debian #550717), sync SUSv4 wording, improve comments

Browse Source
This commit is contained in:
tg
2009-10-15 16:36:27 +00:00
parent 9afaf050d8
commit 578d4e6314
3 changed files with 35 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files

56
funcs.c
View File

@ -25,7 +25,7 @@
#include "sh.h"
__RCSID("$MirOS: src/bin/mksh/funcs.c,v 1.139 2009/10/15 16:32:50 tg Exp $");
__RCSID("$MirOS: src/bin/mksh/funcs.c,v 1.140 2009/10/15 16:36:26 tg Exp $");
#if HAVE_KILLPG
/*
@ -644,7 +644,7 @@ c_print(const char **wp)
for (c = 0; ts[c]; ++c)
Xput(xs, xp, ts[c]);
continue;
}
}
}
Xput(xs, xp, c);
}
@ -2133,32 +2133,37 @@ c_eval(const char **wp)
s = pushs(SWORDS, ATEMP);
s->u.strv = wp + builtin_opt.optind;
/*
* Handle case where the command is empty due to failed
* command substitution, eg, eval "$(false)".
* In this case, shell() will not set/change exstat (because
* compiled tree is empty), so will use this value.
* subst_exstat is cleared in execute(), so should be 0 if
* there were no substitutions.
/*-
* The following code handles the case where the command is
* empty due to failed command substitution, for example by
* eval "$(false)"
* This has historically returned 1 by AT&T ksh88. In this
* case, shell() will not set or change exstat because the
* compiled tree is empty, so it will use the value we pass
* from subst_exstat, which is cleared in execute(), so it
* should have been 0 if there were no substitutions.
*
* A strict reading of POSIX says we don't do this (though
* it is traditionally done). [from 1003.2-1992]
* POSIX however says we don't do this, even though it is
* traditionally done. AT&T ksh93 agrees with POSIX, so we
* do. The following is an excerpt from SUSv4 [1003.2-2008]:
*
* 3.9.1: Simple Commands
* 2.9.1: Simple Commands
* ... If there is a command name, execution shall
* continue as described in 3.9.1.1. If there
* is no command name, but the command contained a command
* substitution, the command shall complete with the exit
* status of the last command substitution
* 3.9.1.1: Command Search and Execution
* ...(1)...(a) If the command name matches the name of
* a special built-in utility, that special built-in
* continue as described in 2.9.1.1 [Command Search
* and Execution]. If there is no command name, but
* the command contained a command substitution, the
* command shall complete with the exit status of the
* last command substitution performed.
* 2.9.1.1: Command Search and Execution
* (1) a. If the command name matches the name of a
* special built-in utility, that special built-in
* utility shall be invoked.
* 3.14.5: Eval
* ... If there are no arguments, or only null arguments,
* eval shall return an exit status of zero.
* 2.14.5: eval
* If there are no arguments, or only null arguments,
* eval shall return a zero exit status; ...
*/
exstat = subst_exstat;
/* exstat = subst_exstat; */ /* AT&T ksh88 */
exstat = 0; /* SUSv4 */
savef = Flag(FERREXIT);
Flag(FERREXIT) = 0;
@ -2331,9 +2336,10 @@ c_set(const char **wp)
for (wp = l->argv; (*wp++ = *owp++) != NULL; )
;
}
/*
/*-
* POSIX says set exit status is 0, but old scripts that use
* getopt(1) use the construct: set -- $(getopt ab:c "$@")
* getopt(1) use the construct
* set -- $(getopt ab:c "$@")
* which assumes the exit value set will be that of the $()
* (subst_exstat is cleared in execute() so that it will be 0
* if there are no command substitutions).