-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Thursday 25 September 2003 09:52, [EMAIL PROTECTED] wrote:
> itself. The thinking is that the quality of the code is much lower; that

and yet closed source software continues to have as many if not more serious 
security issues. so much for "quality" of code...

there's also the issue of having all open source code, well, open. it's easy 
to look at it and criticize it since you have the luxury of seeing it and 
criticizing it. it's much harder to be critical of the quality of code you 
can't see. i think anyone who's worked in real life situations on closed 
projects will tell you that people's ability to write code doesn't magically 
improve just because nobody can see the code; the opposite is true IME. most 
coders suck, therefore most code sucks. period.

> many people do not even review the source with a very scrutinizing eye and

key word: "many". reality: "enough". if it wasn't being scrutinized, then 
patches to Free Software security issues would always lag behind what the 
blackhats knew and were exploiting. but it's the exact opposite. patches for 
security issues affected Free Software are almost always ahead of the game, 
and the ONLY way you can manage that is if there is serious review with a 
"scrutinizing eye" by the good guys.

> those that do don't have the expertise required to pinpoint the various
> vulnerabilities that may be present. 

people in the security industry tend to be a bunch of arrogant assholes. the 
coders think they are hotter than the sun's core, and the salespeople/execs 
usually have little compunction about stretching the truth or speaking 
outside their realm of knowledge just to push sales. that's how we go the 
"virus protection" industry and l33t speak. outside of their actual security 
analysis of specific pieces of software, i don't much weight behind anything 
they say because it's usually wildly innaccurate.

here's the text of a mail i posted to another list last week regarding the 
dual sendmail and OpenSSH security announcements that i think states the 
realities of software security much clearer:

===
well, we always need to be concerned when it comes to security issues. Open
Source is not a panacea; there  are security issues that arise from time to
time and they should be taken seriously. the Internet is a shared resource,
so we each have a responsibility to ensuring its health. it's quite important
how security issues are discovered and handled and what our options as users
are.

with closed source software, the vendor has very little motivation to take of
security issues in a timely manner. this is because fixing a security problem
takes time and money, and there's usually no means to recoup that investment
if you release it as a patch. many closed source companies take care of
security issues primarily to avoid disenfranchising their users. if they feel
they can get away with not fixing it (the public doesn't know, their users
aren't concerned, etc), they usually try and do just that. fortunately, many
vendors of closed source software are more sensitive to this issue today
since Linux and other Free Software projects have pushed security to the
forefront; we still see far too much lip service being paid, however.

in the closed source world, threats are usually discovered by the black hats
and exploited long before the users are informed about it. the vendor may or
may not know, but when they do they often they keep it under their hat hoping
no one will find out before their next scheduled upgrade (which they can
charge for), if at all. but as a user, you never really know. anybody
remember that MS security bulletin last year warning about massive intrusions
ocuring that they "couldn't explain" but felt the need to warn admins about
anyways?

with Free / Open Source Software, there is a real motivation to find problems
and fix them: the code is open for anyone to peruse (including the black
hats) and those who work on the software usually depend on it themselves as
users. we also have the benefit of the "many eyes make bugs shallow"
principle, which usually works pretty well for security. but we also have
choice: what to use, when and how. this is a bigger ally than it's often
given credit for.

in the case of the sendmail exploit, more and more people are (thankfully)
using something else, such as Postfix. for those who are stuck with sendmail
for one reason or another, more's the pity, but fortunately for them patches
are readily available. they could also switch to another MTA temporarily (or
permanently) at no cost beyond that of installing and starting the
alternative MTA, while waiting for binaries.

but for many this sendmail flaw isn't an issue because we have a plethora of
choices: postfix and (the non-Free) qmail for example; more Linux
distributions are shipping with postfix as a default rather than sendmail
these days. is that sort of inexpensive, easy and pragmatically driven
choice typically available in the closed source world? nope. when Exchange
has a flaw, Exchange has a flaw and you just have to hope that the fix comes
out from MS sooner rather than later (it's often already later by the time
you know about it, however). switching in the meantime is expensive and
laborious, if at all possible.

as for the OpenSSH flaw, as far as I'm aware all major Free Software operating
system vendors have already shipped updated packages. there are still
questions as to the severity of the flaw: it's definitely DOS'able, but
perhaps not fully exploitable (the buffer is on the heap which makes it
trickier). the fact there is debate means that there aren't any (widespread?)
exploits in the wild ... which means that this is yet another non-starter for
those who would like to demean Free Software.

i know that all my systems, except one, were updated automagically as the
patches became available. that one exception is running an older OS version
(i really need to upgrade it, but it's just so stable =) and i had compiled
OpenSSH from source on it; in that case i simply grabbed the source patch and
applied it locally on that machine and recompiled. try that on a
now-unsupported closed source system.

while this was all happening i turned SSH off on the firewalls (one command,
or a couple of clicks, depending on your taste) for the brief period of time
between the announcement and the time the patches trickled down the pipe.
once that happened i flicked SSH back on.

so i haven't been worried at all about the security of my systems over these
issues. and that's a priceless peace of mind and reliability i just don't get
with closed source software
=====

- -- 
Aaron J. Seigo
GPG Fingerprint: 8B8B 2209 0C6F 7C47 B1EA  EE75 D6B7 2EB1 A7F1 DB43
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2-rc1-SuSE (GNU/Linux)

iD8DBQE/dG3m1rcusafx20MRAo+8AKCR1LIRTF1h1joZq0MEsISEm1oRmgCgpdOR
CPpvO3TTaE7zjH6N/Ytgh7g=
=GTFG
-----END PGP SIGNATURE-----

Reply via email to