[racket] Hosting the try-racket REPL.

That does sound an awful headache for something that otherwise doesn't seem
to be using X for anything other than dependency expectations. It seems
like there should be a non-X-dependent way to do it.

On Mon, Sep 22, 2014 at 9:52 AM, Neil Van Dyke <neil at neilvandyke.org> wrote:

> Amir Ansari wrote at 09/22/2014 02:28 AM:
>
>> Have you tried Xvfb (http://en.wikipedia.org/wiki/Xvfb), a virtual X
>> server? It runs headlessly, sidesteps the whole issue of having to fork the
>> code...
>>
>>
> I have no idea whether "xvfb" is the best way without looking closely at a
> particular case, but I've seen "xvfb" work in the past on production
> servers.  (Not for Racket, but for some R statistics tool that touched the
> X server, so that the tool could be used off-the-shelf, rather than
> maintaining a modified fork of the tool source.)
>
> If you do this, just be write your startup and shutdown scripts to be
> resilient and stable (e.g., wrt available ports/sockets at the time, and to
> shut down cleanly, and so there are no race conditions or conflicts between
> multiple simultaneous users), and secure (e.g., ensure that it won't
> accidentally listen on a public network interface, and that an attacker
> can't synthesize operations to the xvfb to escalate privileges or DoS, and
> that the xvfb can't be used to leak data between users).
>
> Neil V.
>
>
> ____________________
>  Racket Users list:
>  http://lists.racket-lang.org/users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.racket-lang.org/users/archive/attachments/20140922/d040ea62/attachment.html>