[PHP-DEV] Re: [RFC] Improve openssl_random_pseudo_bytes()

2018-10-31 Thread Sammy Kaye Powers
Hey Tom! > According to one argument that has a lot of currency, it does. You have great points that I totally agree with; after all, my original proposal was to alias to random_bytes(). But this RFC just targets the implementation problems, not the OpenSSL CSPRNG itself. We already discussed in

[PHP-DEV] Re: [RFC] Improve openssl_random_pseudo_bytes()

2018-10-22 Thread Tom Worster
Hi Sammy, On 22 Oct 2018, at 9:46, Sammy Kaye Powers wrote: What makes the function obsolete? The addition of the `random_bytes()` Yes. What makes the function obsolete? The addition of the `random_bytes()` CSPRNG (which uses the kernel's CSPRNG) doesn't invalidate OpenSSL's CSPRNG. Accor

[PHP-DEV] Re: [RFC] Improve openssl_random_pseudo_bytes()

2018-10-22 Thread Sammy Kaye Powers
Hey Tom! Thanks for the feedback! :) On Sun, Oct 21, 2018 at 8:09 PM Tom Worster wrote: > > At first glance I believed you were proposing that > `openssl_random_pseudo_bytes()` should fail with an exception and that this > would be an improvement. I would agree with that. Yep! If the function

[PHP-DEV] Re: [RFC] Improve openssl_random_pseudo_bytes()

2018-10-21 Thread Tom Worster
On 19 Oct 2018, at 16:46, Sammy Kaye Powers wrote: I'd like to start a discussion on the "Improve openssl_random_pseudo_bytes()" RFC: https://wiki.php.net/rfc/improve-openssl-random-pseudo-bytes TL;DR: CSPRNG implementations should always fail closed so this change would make `openssl_random_p