[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <56FDB478.4020609@infradead.org>
Date: Thu, 31 Mar 2016 16:36:24 -0700
From: Randy Dunlap <rdunlap@...radead.org>
To: Richard Weinberger <richard@....at>, linux-kernel@...r.kernel.org
Cc: linux-api@...r.kernel.org
Subject: Re: New syscall: leftpad()
Please be more careful in your description...
On 03/31/16 15:33, Richard Weinberger wrote:
> Recent happenings in the node.js community showed how fragile software is when
> it comes to dependencies of fundamental algorithms like leftpad[1].
> A node.js package which provided ledpad vanished and broke a lot of software.
leftpad
> This raised our attention and we came to the conclusion that it is the kernel's
> job to provide such functionality such that node.js based applications can in future
> rely in Linux's "don't break userspace" rule.
> We hope that glibc and Andoid's bionic will soon offer wrapper functions for this
Android's
> new leftpad system call.
> We put leftpad into the kernel not only because of Linux's stable ABI,
> also for performance reasons.
> As everyone knows, within the kernel everything is faster and better.
> Leftpad has millions of users, so it has to be as fast as possible.
> This new system call will also help making services like left-pad.io[2]
> faster and more reliable. If the leftpad() system call gets adopted by a wider user base
> it might also make sense to add a generic npm() system call which acts like ioctl()
> where kernel modules can register new functions that are often used by node.js.
> Such functions might be, is_array(), is_int(), etc.
--
~Randy
Powered by blists - more mailing lists