Jeremy Evans <jer...@openbsd.org> writes:
> On 11/04 11:19, Jeremie Courreges-Anglas wrote:
>> Jeremy Evans <jer...@openbsd.org> writes:
>>
>> > This is based on the ruby-tame port I worked on at c2k15, which
>> > apparently was too early. Using this, you can access pledge(2)
>> > from ruby via:
>> >
>> > require 'pledge'
>> > Pledge.pledge('rpath') # stdio added automatically, as ruby needs it
>> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>> This bugs me, since "stdio" is needed by pretty much any program. the
>> following program dies with SIGABRT:
>>
>> #include <unistd.h>
>>
>> int
>> main(void)
>> {
>> if (pledge("", NULL) == -1)
>> return 1;
>> return 0;
>> }
>>
>> Why would the ruby interface be different?
>
> Pretty much any program is not all programs. It is possible for a C
> program to run without the stdio promise, if it purely operates on
> shared memory and just calls _exit(2) when it wants to exit, as
> specified in the man page.
Duh, I had completely missed that bit. I must admit it's been a long
time since I've last looked at the manpage. My bad.
> It is impossible for a ruby program to run without stdio, there are
> various things ruby does internally that require stdio.
>
> Also, good ruby APIs would not require you to explicity specify
> something that is always implicitly required. Ruby is not C.
I see, please ignore my initial concerns.
> Thanks,
> Jeremy
>
--
jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
