
joyce.hopewell at gmail
Nov 2, 2007, 1:29 PM
Post #5 of 14
(1514 views)
Permalink
|
|
Re: "Some thoughts on security after ten years of qmail 1.0"
[In reply to]
|
|
Emilio Perea wrote: > I just noticed Dan has a new paper on qmail and security: > > http://cr.yp.to/qmail/qmailsec-20071101.pdf It's astounding how useless that 'paper' is. It's nothing more that a self-aggrandizing plea to the Internet at large to not forget that DJB still actually exists. Here's how I see it. qmail has its place, sure. However, it has not kept pace with the rest of the Internet. It's an anachronism. To get your qmail platform to use any sort of modern e-mail capabilities (greylisting, spf, dkim, etc) you have to apply a mountain of patches, piped scrips, helper programs ... it turns into a giant uncontrollable mess very quickly. Yes, qmail is fast. Yes, it's arguably more secure than sendmail (altho, if I recall correctly, DJB has welched on his $500 bet in the past. The 32/64 bit controversy, wasn't that?). However, by forcing users to jump through a thousand hoops to add functionality, it stops being secure or fast, or reliable. So, either you have a marginalized stock qmail install, or an unwieldly shoehorned qmail install. Either option is very unappealing. I really think this paper is nothing more than an attention grabber. A desperate attempt to say, "I still exist!". But honestly, qmail has way too many shortcomings to be a serious contender in any modern datacenter anymore. I've worked with qmail for a long time. I know it very well. I know you can add any sort of support for it with patches or pipelines or proxies. But I only use it because I have to support legacy installs. I refuse to use it for any new installations, or any personal platforms. Face it, qmail is abandonware. Will there ever be a qmail2? I really don't care. There are other packages available that give me the functionality I want now, not years from now. And for heaven's sake, I'm sick to death of the whole errno fiasco. The world moved on. Face up to the facts, just rewrite the header files already, and release qmail-1.03.1. At least then we could pretend qmail was still relevant.
|