On Mon, Apr 30, 2007 at 08:36:24PM +0200, Henrik Brix Andersen wrote: > Todays changes to *env() appears to have broken stuff (my -CURRENT box > fails to build many ports after these changes, due to missing > environment variables in the build environment). > > Here is a simple test-case: > > $ env -vi foo=bar env > #env clearing environ > #env setenv: foo=bar > #env executing: env > #env arg[0]= 'env' > > Notice that the second instance of env(1) doesn't see the $foo > environment variable. > > This is the output of the same test on my RELENG_6 box: > > $ env -vi foo=bar env > #env clearing environ > #env setenv: foo=bar > #env executing: env > #env arg[0]= 'env' > foo=bar
Thanx for the report, but please be patient a bit, work in the progress
and fix already planned soon.
To the matter: env incorrectly calls setenv("name=...", ...) which is
POSIX violation:
"The setenv( ) function shall fail if:
[EINVAL] The name argument is a null pointer, points to an empty string,
or points to a string containing an '=' character."
--
http://ache.pp.ru/
pgplTIXaXLSGF.pgp
Description: PGP signature
