On Tue, 2005-03-01 at 11:31 +1000, QuantumG wrote: > The C language is just so full of pitfalls and traps that it is next to > impossible to write secure code in it. You can do it obviously (look at > OpenBSD) but it requires so much discipline and arcane knowledge that > most programmers simply can't write secure code.
Secure programming howto and the articles by David Wheeler on Alpha works. These are eye opening, ever considered a temp file named mytemp$$ is a security hole? http://www.dwheeler.com/secure-programs/ http://www-106.ibm.com/developerworks/linux/library/l-sp1.html > The number of people reporting exploitable bugs in software has dropped > in the last 5 years. It's not because the software has gotten better. I cannot believe this. Certainly there is things that are just not fixed however somethings are becoming more secure the more eyes that are upon it. Also the advent of code checkers is discovering some of these exploits without intensive code review. There are always new exploits discovered and there are people who do University courses discovering those exploits the results are published. It is up to the projects to then take the problem seriously that can be difficult. Firefox is security conscious because it is being attacked but does terminal software require the same level of intense scrutiny? What would happen if I came up with an exploit in gnome-terminal would it be fixed immediately? It really does depend on the application. I have to say the kernel is a rich source of exploits because: a) it is huge with lots of options. b) cracking it gains much more access. c) There are chunks of it not closely maintained, ie the drivers. -- Ken Foskey OpenOffice.org developer -- SLUG - Sydney Linux User's Group Mailing List - http://slug.org.au/ Subscription info and FAQs: http://slug.org.au/faq/mailinglists.html
