D. J. Bernstein

http://www.aaronsw.com/weblog/djb

by access_denied • 17 years ago

107 116 17 years ago

aaronsw.com

jacquesm3• 17 years ago

I think he would have my vote, but there are some other 'great' programmers out there.

how about this for minimalism: http://cr.yp.to/

tzury2• 17 years ago

man, this is the man! http://cr.yp.to/djb.html

owinebarger• 17 years ago

Despite the purple prose, I generally agree with Aaron's sentiments about great programming ("the best", I can't say, but definitely "great"). He is, however, also very rude, at least in his online persona. Having "a forceful, uncompromising vision" is an orthogonal issue to how you choose to communicate.

idlewords2• 17 years ago

I can't read anything Aaron writes without my mind flying back to his seminal essay on the unanticipated softness of the human breast:

http://them.ws/post/1736/testing_something

ErrantX5• 17 years ago

Thanks for that. Not come across it before and I cant remember a blog post that actually touched me so deeply!

niels• 17 years ago

Not to forget daemontools.

dustingetz• 17 years ago

I was surprised and looked him up on wikipedia -- he did indeed write daemontools.

randallsquared• 17 years ago

daemontools and tinydns are far, far better examples of brilliance than qmail, in my opinion. Especially tinydns comes far closer to meeting modern needs than qmail, for its domain. I'll never go back to the horror of bind config files.

bd• 17 years ago

Not to be confused with Daemon Tools :)

http://en.wikipedia.org/wiki/Daemontools

http://en.wikipedia.org/wiki/Daemon_Tools

whirlycott12• 17 years ago

I have dealt with qmail for ten years. It is a complete joy once you have it up and running. You literally can stop thinking about it because it always works. However, getting it up and running requires adding in antispam, antivirus, virtual mailboxes, some additional protective and administrative patches, some extra stuff for SMTP-AUTH, SSL/TLS, reporting, and perhaps a few more patches for fun.

Patches, you say? Well, I'm glad you really like source code, because that's exactly what you'll be dealing with to get all of this up and running. I've done it many times before manually and it takes around ten hours to do it.

While Aaron's point is that djb's code, vision and discipline are possibly unmatched (I concur on all counts), the software (principally qmail) is antiquated in the sense that nobody needs just that one part in order to run a semi-modern email system.

So, as you start downloading the .tgz for qmail and begin studying it like good campers, remember that this wonderful code is, by itself, not up to the job that it purports to serve.

-- Lots of love,

Someone who currently runs two qmail systems and will never go to the trouble of setting one up again

andreyf4• 17 years ago

If code I write is hard to deploy to your system, I don't think my code can be to blame. Rather, it's the fault of the system/language in which we are working.

astine4• 17 years ago

I hate this attitude. If your code targets my system, I should be able to install it without editing it.

jrockway4• 17 years ago

I agree with you. DJB wrote secure code that works very well and donated it to the public domain; if you want it to be Perfect For Your Needs, you are going to need to do that yourself. You are a unique snowflake, and your needs differ from DJB's.

"You suck for not doing enough work" is no way to treat someone giving you free software.

mseebach2• 17 years ago

> I've done it many times before manually and it takes around ten hours to do it.

This is exactly why I've been using Postfix (or, preferably, get Google or someone else to handle it for me).

Easy to install is a feature, and a pretty close cousin of the "shipping" feature.

DJBs software is interesting mostly from an academic perspective. In the real world, it's hard to justify using something else, when apt-get gets the job done in literally seconds.

alec2• 17 years ago

I haven't used qmail, but I use djbdns. The complete installation, possibly including a few patches, is less work than figuring out how to set up BIND's configuration. It's available in Debian from apt-get.

It's good that it's so easy to figure out how to use, because I touch it less than once a year, so I've completely forgotten about it. That's great software.

wallflower• 17 years ago

Bill Joy

"As a UC Berkeley graduate student, Joy worked for Fabry's Computer Systems Research Group CSRG in managing the BSD support and rollout where many claim he was largely responsible for managing the authorship of BSD UNIX, from which sprang many modern forms of UNIX, including FreeBSD, NetBSD, and OpenBSD. Apple Inc. has based much of the Mac OS X kernel and OS Services on the BSD technology.

Some of his most notable contributions were the vi editor, NFS, and csh. Joy's prowess as a computer programmer is legendary, with an oft-told anecdote that he wrote the vi editor in a weekend. Joy denies this assertion.[2]"

http://en.wikipedia.org/wiki/Bill_Joy

rit• 17 years ago

Exactly my thought when I read this. IIRC he wrote one of the early TCP/IP implementations as well, which a lot of OS' still use a derivative of. And dns and mail rely upon.

The wikipedia article links specifically to a sadly abandoned project that Salon had years ago to document the free software revolution.

http://dir.salon.com/story/tech/fsp/2000/05/16/chapter_2_par...

I've always remembered it and wished they'd finished but it sadly seems to have stopped getting new entries in '01:

http://dir.salon.com/topics/free_software_project/

nailer5• 17 years ago

I should add NetApp's internal OS is also BSD based - ironic since NetApp and Sun have been at each other's throats for years.

tzury2• 17 years ago

Linus, RMS, Guido, and others. I can think of so many other names.

You know what, even John Resig's jQuery software is served by millions of servers, running in billions of web pages every day.

This all "THE BEST" thing, is somehow pointless IMO.

mikegioia• 17 years ago

I have to agree with you on JQuery. It runs on multiple browsers too and performs exceptionally well. It's gotta be one of my all time favorite pieces of software.

chaosmachine• 17 years ago

It's hard to overstate how great jQuery is. I'm always surprised at how easy it makes things. I remember the bad old days, spending hours researching the different ways to reference dom objects, trying to find a method that worked in more than one browser. With jQuery, you really can just code it once, and it works everywhere.

davidw• 17 years ago

It really does seem a bit pointless. Why not just say "djb writes great, beautiful code - some of the best I have ever seen" and leave it at that?

rortian2• 17 years ago

I agree, but aaronsw must come at an issue as a raging partisan. I know that in economics, philosophy, politics, and apparently best programmer ever!!!, he always comes at the issue with a deeply contentious point of view from an entrenched idealogical camp.

It's kind of a shame because of the the advantages of being young is being able to approach a field without having to take sides. Meaningful dialog and inquiry void of motivated reasoning are beautiful things. Unfortunately, some people need an argument.

twism• 17 years ago

jQuery gives me goosebumps whenever I get a chance to work with it.

aristus• 17 years ago

You can turn that off:

    $.fn.defaults({goosebumps: false});

nailer5• 17 years ago

Only two others would equal DJB in terms of having more than one major well known app (a mail server and a DNS server). In my experience, I wouldn't say djbdns is particularly widely deployed...

Linus Torvalds

* The Linux Kernel

* git

Fabrice Bellard

* FFMpeg (if you've done anything with video in the last 10 years, you've used this)

* QEmu (basis for QEmu emulator and userspace portion of KVM VM). Ubuntu's open EC2-API compatible elastic cloud app uses this, as does Red Hat Advanced Platform.

If there are any others, let me know. Mongrel's pretty popular, if Lamson does the same maybe we could add Zed.

pilif• 17 years ago

This is slightly off-topic as it's not about djb, but about qmail. Still:

I have some experience with qmail. While I agree that it's generally very nice and certainly incredibly bug free, that quality comes at a price: Lack of features.

The moment you start adding stuff: Server-side filtering (like sieve), LMTP, virus checking, spam filtering, virtual users - all that stuff you need to run a modern mail server - you will have to patch qmail.

And by manually patching your installation, you are picking up responsibility: Now that you are not running stock qmail, but something you patched and installed yourself, you will no longer be able to rely on your distribution for security fixes and you'll have to keep updated and repatch yourself.

And while qmail is beautiful and bugfree, the same can not always be said for the additional patches, so you WILL be patching.

If you get all the patches for the functionality you need to actually play together that is.

If qmail solves your issue and you are prepared for this, then go with qmail. If you need a ton of functionality built-into your mailserver, probably go Exim and if you need good compromise between featureset, architecture and security, then you'll probably take postfix.

tptacek10• 17 years ago

Please avoid Exim. Even if you can't stomach qmail, which is an acquired taste, you don't have to compromise on security. Smart people can disagree about whether Postfix is _as_ secure as qmail, but none of them disagree that it is at least _very_ secure.

mhd• 17 years ago

What is this, the Hill Valley International Programming Tournament?

People have arguments like this all the time in the sports sector, and at least there you can come up with all kinds of statistics and recordings, but what do you have for programmers? If they produced Open Source, you might have the end result of their work, but that would exclude a large part of the work force. And even then, what are your points of measurements? Just the amount of bugs? Total? Per year? Per line? Meh.

And as a final, somewhat unrelated note: Which I have nothing against Mr. Bernstein himself, I've got a pretty low opinion of most djb fanboys I've encountered in the past...

antirez6• 17 years ago

(read this message with a background of "arguing about the best programmer is anyway silly")

djb is in no way the best programmer of the world, but he is a very smart programmer, and probably an even better mathematician. And now, my arguments about why he is not the best programmer.

1) djb software is not the most elegant software around at all. Actually trying to write high performance and bug free software his style is the most procedural possible, with very little abstractions layers.

2) his software is well known to be uneasy to configure. There are qmail hackers around. When there are <name of a unix deamon>-hackers around it's a bad sign about the quality of the interface between the program and the rest of the system and the system administrator itself.

3) Djb played no role at all in the programming languages world. AFAIK he never suggested some new idea for programming languages or new directions that are now considered important. You can't be the best without being influential in the field.

4) Djb newer wrote big systems that were crucial for the programming world. Qmail and djbdns can disappear tomorrow and everything will be up and running anyway. GCC or the Linux Kernel are a different matter.

I can continue, but I think 1+2+3+4 are already enough to deny the argument aaronsw is trying to push in the article.

alec2• 17 years ago

"You can't be the best without being influential in the field."

I think an appropriate analogy is judging an architect on the hammers he has improved instead of the buildings he has designed.

antirez6• 17 years ago

I disagree, the buildings designed by Wright or Le Corbusier improved the field in many ways. They created a new architectonic system, it's like inventing a new programming language in some way.

access_denied• 17 years ago

aguing about the best programmer is a smart move when seen as an exercise to find out what it is that makes a programmer good.

antirez6• 17 years ago

this is a good point, but can be expressed with a blog post about "what are the qualities to be among the top programmers of the world?" instead of "djb is the best programmer of the world".

Still it is surely important to look at this great programmers in the aim to try to emulate all the stuff that they got right in order to improve ourself.

crucini• 17 years ago

I question #3. Why is it admirable to play a role in the languages world? Shouldn't languages be a means to an end? If the end is good applications, DJB showed that you can produce good applications in C.

It seems like a lot of the energy around languages is frivolous. It's more fun to play with language features than to attack a real-world problem like writing a mail server.

As for #4, why does criticality matter? And why does the audience have to be programmers? Couldn't the author of a spreadsheet be as worthy as the author of a compiler?

Not that I'm endorsing DJB's nomination; I have pored through his source code for various reasons and have some reservations about it. But you could do much worse.

(And yes, it's all silly.)

antirez6• 17 years ago

#3: sure, in general terms. Not true in my opinion for the best programmer of the world. Naturally he is "the one" that can master programming so well that it must be able to drive even programming language designers into more expressive / productive languages. djb can never be the one as in his code is very rare to find high level abstractions.

as for #4 it's very admirable to write real world software, but the best programmers in the world, like Joy or RMS tend to write big systems that can be used by the other hackers in the world, because to write a new operating system, a C compiler that live for decades contributing even to the development of new operating systems (It's hard to imagine Linux without GCC in some way), or world class text editors (vi, emacs) is not something everybody can do, but only the best programmers in the world.

there• 17 years ago

your points 2 and 3 directly conflict with each other. his software is "difficult" to configure because he does it in a non-standard way. individual files for options, installing everything in its own root directory, using /service for daemontools, etc.

he had these new ideas for software development, installation, and maintenance, so he implemented them and because they are different, people equate that with being difficult. he didn't like a lot of standard libraries, so a lot of his software uses his own routines (which has been extracted as libdjb - http://www.fefe.de/djb/) which is partially the reason for his impressive security record. i think you underestimate his impact on the unix software development community.

your fourth point is just wrong. a lot of ISPs rely on qmail and djbdns and to say that they wouldn't care if they just disappeared is just silly. sure there are alternatives, but so are there to linux and gcc.

techiferous• 17 years ago

"One bug — one bug! — was found in qmail."

While this is certainly impressive, the number of bugs is not a good metric for measuring the effectiveness of a programmer. A programmer is only as good as the business problems she solves.

One bug in a life-critical application could be catastrophic. On the other hand, a programmer that makes sure their code is completely bug-free would not do well in a domain where time-to-market is critical and quality of service is not.

idlewords2• 17 years ago

Also, the number is at least four bugs. http://cr.yp.to/qmail/qmailsec-20071101.pdf

antirez6• 17 years ago

and this is about just security bugs. All the other bugs are not counted.

ErrantX5• 17 years ago

It's a lovely tribute to djb and you cant really fault it.

Except to say "best" is such a subjective thing as to make it useless.

There are many programmers deserving of a "hall of fame" entry.

l0stman• 17 years ago

> No other programmer has this kind of track record. Donald Knuth probably comes closest, but his diary about writing TeX (printed in Literate Programming) shows how he kept finding bugs for years and never expected to be finished, only to get closer and closer (thus the odd version numbering scheme).

I don't want to be dismissive but I think TeX is more complex and uses non trivial algorithms. So it's like comparing an apple to an orange.

jacquesm3• 17 years ago

Anybody that says an email server is 'non trivial' hasn't looked at the guts of one yet.

tptacek10• 17 years ago

As someone who has hacked mail servers, written DNS servers, and come within 100 miles of typesetting, I'm going to throw in my vote that TeX addresses a much harder problem than qmail.

rv77ax• 17 years ago

Dennis MacAlistair Ritchie & Kenneth Lane Thompson.

thread closed.

known• 17 years ago

Bill Gates

ErrantX5• 17 years ago

I think you're being a little unfairly downvoted. I don't think anyone would claim Gates should win prizes for minimal or beautiful code.

But he certainly did a lot of good programming.

In any hall of fame where we mention Linus, Guido and Bernstein we probably do need to mention Gates as well.

viraptor2• 17 years ago

And where can we find any of his programs to look at?...

blasdel• 17 years ago

What, you think DONKEY.BAS is exemplary code?

nailer5• 17 years ago

What well known software did Bill Gates write? DOS was purchased from another company. The oldest parts of Excel maybe?

pg• 17 years ago

He wrote Microsoft's original Basic, probably parts of other languages. DOS must have been changed a lot after Microsoft bought it. He may have written some of the new stuff.

allan_• 17 years ago

This article is pathologically sick. Couldn't read it any more, to painful. qmail gives you pain, too. It is completly useless without applying patches, which are mostly of poor quality. And it misleads you to do most unperformant things around it. Like writing shellscripts which fork() like hell for every mail there is to process.

kilowatt• 17 years ago

I was with Aaron until

> What other field combines all these arts? Language, math, art, design, function. Programming is clearly in a class of its own. And, when it comes to programmers, who even competes with djb? Who else has worked to realize these amazing possibilities? Who else even knows they are there?

Is this satire?

unalone2• 17 years ago

Yeah, only slightly obnoxious. Nothing worse than somebody so relatively new to a field that they still think it objectively is better/cooler/more fulfilling than every other field.

aaronsw4• 17 years ago

It's hard to claim I'm new to programming. I did it for over a decade and then retired from it. Perhaps it's just nostalgia, then.

ecq• 17 years ago

the best programmer is the person who have made the biggest difference in people's lives.

it's not measured by the no. of lines of code, bugs,etc.

capablanca• 17 years ago

So, Bill Gates.

gjm113• 17 years ago

Biggest positive difference.

c00p3r• 17 years ago

There are several less known project with really passionate coding, especially http://nginx.net/. postfix, sqlite and many other projects could be compared with qmail.

mgrouchy• 17 years ago

who cares?

csmajorfive• 17 years ago

I love djb and his tools but, as someone who hacked djbdns extensively for a research project, I think the project misses empathy for other programmers. It's hard to follow, poorly documented, and too clever sometimes. That said, if we judge code only on performance, then yeah it's rock solid.

aaronsw4• 17 years ago

I much preferred the old title. Can a moderator put it back?

mikeryan• 17 years ago

I have a feeling that a large portion of the best programmers work in absolute obscurity.

I worked for a long time with guys that wrote embedded software. Absolutely brilliant guys that never get their due because no one knows how they were.

tptacek10• 17 years ago

DJB's code is a lot of things, but when someone says "beautiful" is one of them, I start thinking about how I might quiz them to get them to prove that they've actually read it. Smarter coders than me have sat in rooms for nightlong studies of qmail and come to the conclusion that, while clever, the C code in qmail and djbdns has clearly been compiled down from some higher level language[1]. If that code is anything, it is idiosyncratic.

[1] Having asked this question directly to DJB in person, I can say that I am at least convinced he wrote this stuff in C.

andreyf4• 17 years ago

Having so clear a vision of how your architecture should behave that you human-compile it into idiosyncratic code that looks generated seems to be a very good definition of beautiful code, no? Maybe he hasn't had time to write a book about the "code patterns" which he's using, but that's hardly something to hold a grudge about.

silentbicycle• 17 years ago

If you write code that looks like it was generated with e.g. yacc, then you probably would have been better off writing it in yacc and bundling that as the source. There's nothing glamorous in being a human compiler. Other people will have to touch the code at some point.

I haven't read the qmail source (just some design docs about how the whole system fits together, which I found rather impressive), so I'm not talking specifically about that.

NB: Qmail's license wasn't public domain until 2007. (Also, damn, tptacek, you're fast! I deleted that part, since I'm talking about auto-generated code in general.)

tptacek10• 17 years ago

Especially if you're a security person, I think it's uncontroversial to say that the design is beautiful. It's not only an elegant design, but it is the first major piece of systems code to take that design approach. To put it differently: most of major systems programs that have been proven secure in the last 10 years owe their design to qmail.

But that has nothing to do with the code, which is not only epsilon from assembly (Bernstein fully embraces the notion of writing code in high-level assembly), but also clever and concise almost to a fault.

As someone who ran qmail since it was originally released in beta, I also remember vividly Bernstein's original idea about configuration, which is that "configuring" your mail server with C code was more reasonable than learning another programming language (Sendmail "cf"). Which implies that a lot of the code in the interesting parts of qmail are less about design, and more about encoding mail routing policy as C code.

wheels• 17 years ago

Having read some of DJB's code as well, I also thought, "Huh?" when it was suggested that it was beautiful. One thing that would lead a person to believe the compilation argument is the almost complete lack of comments, as-short-as-possible variable names, etc.

tptacek10• 17 years ago

The formatting, the heavy reliance on idiosyncratic CPP macros, and more than anything else the repetitively procedural nature of it is what set off alarm bells for us.

But really that's just the way he codes. It's very intricate. It's like a very ugly Swiss watch.

aaronsw4• 17 years ago

Are you sure he wrote all of it in C? He often talks about higher-level languages he's written.

tptacek10• 17 years ago

Can I promise you that there's no codegen in it? No. Would I bet real money that there's no codegen in it? Yes.

shawnps3• 17 years ago

Could someone clarify this for me, please? On this page:

http://cr.yp.to/djbdns/intro-dns.html

under the "Multiple Servers" section, he says

"To protect against computer failure, there are actually several root servers, several .to servers, and two yp.to servers."

I don't understand, is he saying there are only two DNS servers in the world that you can contact to resolve yp.to?

aaronsw4• 17 years ago

Yes, that's true for most domains. All the other servers have to cache responses from those two servers. (For the record, the two yp.to servers are currently djb's machines dancer and forcewall.)

shawnps3• 17 years ago

What would happen if he suddenly decided to unplug dancer and forcewall forever, would all the other DNS servers in the world fail to resolve yp.to after their cached responses expire?

shawnps3• 17 years ago

Is there a way to find out for any given domain how many actual servers hand out responses to other DNS servers to hold in their caches?

wmf• 17 years ago

You can use WHOIS or dig, e.g.

$ dig yp.to ns

; <<>> DiG 9.4.3-P3 <<>> yp.to ns ;; global options: printcmd ;; Got answer: ;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 63068 ;; flags: qr rd ra; QUERY: 1, ANSWER: 2, AUTHORITY: 0, ADDITIONAL: 0

;; QUESTION SECTION: ;yp.to. IN NS

;; ANSWER SECTION: yp.to. 184700 IN NS b.ns.yp.to. yp.to. 184700 IN NS f.ns.yp.to.

nailer5• 17 years ago

dnstracer can show you that and much more. A great tool, shame nobody's heard of it.

bayareaguy• 17 years ago

djb is also the author of one of my favorite database libraries: http://cr.yp.to/cdb.html

innoncent• 17 years ago

djb wrote the code that won the EngineYard SHA1 challenge. His code was processing 800 million sha1 hashes per second. He wrote CUDA and C implementations. His sha1 implementation was 12 times faster than OpenSSL.

D. J. Bernstein

Related Stories