Re: lilypond via web interface: security considerations

[Mike Blackstock]

Actually we're in agreement; as you put it, a %100 secure system is beyond the means of

most people.

 

So I'll 'rephrase what I wrote earlier : "Although theoretically possible, real world considerations

are such that a %100 secure publicly- accessible webserver that is within the financial means of most

website owners is not possible and nothing will stop a committed hacker who wants access to

your system". Sorry for the confusion.

 

Now that you mention it, though, what I meant was that context is everything. Wikipedia can ill-afford to

deploy Lilypond for good reasons it has given, but those reasons needn't apply to people like me and

others who have expressed an interest in deploying lilypond on a webserver on an experimental basis.

I think somebody with modest-to-good linux skills can deploy a system that is reasonably secure and

by that I mean it can escape detection by the bot scripts that scour the net for trivially hackable systems.

As long as it's not on a public server - and by that I mean one that contains other websites and info -

I think it's doable. I know I tried it on a small basis; I had some pople try to hack it and they said stuff

like 'I don't know how much time you want me to  spend on this but the standard kiddie stuff can't get through".

I ran it in a jail, removed anything that might be useful to someone who even gained root access, I had

Tripwire installed to monitor all files, I had the replacement libc libraries installed (the ones that monitor

for attempted bufferoverflow exploits) and so on.

 

So I think someone who wants to experiment with it shouldn't be dissuaded by the reasons given by

the people from Wikipedia. We should deploy it experimentally, ensure nobody can use the systems to

disrupt other systems and - here's the key part - learn from it.

 

My beer's getting warm - Cheers,

Mike

 

 

 

 

 

 

On Fri, May 22, 2009 at 8:43 AM, Graham Percival <address@hidden> wrote:

On Thu, May 21, 2009 at 02:47:54PM -0400, Mike Blackstock wrote:
> Of course, 'security' is relative - nothing will stop a commited
> hacker who's targeted your system, so I'm a bit mystified by
> some of the other responses here.

That's not true -- The only reason that computer security is a
joke is that people *treat* it like a joke.  It's entirely
possible to create a system that will foil a committed hacker.

Now, does this involve a lot of work?  Certainly.  And consumers
willing to pay to have this work done?  Definitely not!  But that
doesn't mean that computer programs are some magical black box
that anybody can break.  Barring random bit-flipping from solar
rays, computers are deterministic objects.


By "a lot of work", suppose that all programmers (and academics)
stopped implementing new features and new programs in 1994, and
spent the past 15 years just improving security.  How many holes
do you think would be left in the result?

Cheers,
- Graham