Slashdot Mirror
20 Network Changing Products
An anonymous reader wrote to mention a Network World piece about products that have changed networking over the last twenty years. From the article: "SendMail 1998 - Sendmail was key to the e-mail revolution because it was how everyone got up and running with e-mail communications over the Internet. Eric Allman wrote the original version of this open source mail-transfer agent while he was at the University of California at Berkeley in 1979. He stopped development on it in 1982, however, and didn't revisit it until 1990. In 1998 he founded SendMail to sell the software's first commercial version, the SendMail switch."
And Sendmail also happens to be one of the absolute worst widely-deployed programs in the history computer software. Man, I despite that program. How could anyone have thought that configuration file format was a good idea? You know it's bad when you have to have a preprocessor to translate something (semi-)tolerable into its syntax.
The e-mail revolution succeeded DESPITE sendmail, not because of it, though I give it some small credit for flexibility. It was just barely adequate enough to keep people from writing a replacement (thought we have some now).
No point to this post, except to voice how much I despise sendmail. :)
Sometimes it's best to just let stupid people be stupid.
From TFA:
Skype
2003
This proprietary peer-to-peer telephony application provided the first real quality VoIP product (did we mention it's free here?) that has built a cult following and spurred industry questions about why corporations can't move to convergence more quickly. Skype picked up both business clout and deep pockets when eBay bought the company in the fall of 2005
Hello? Asterisk anybody?
Open source? Check
Open standards? Check ( note: skype is not open in this regard )
Quality product? Check check check
Huge business impact? Check
Not to mention asterisk isn't burdened with weird restrictions fueled by marketing concerns. Digium is the company behind it, and they do make hardware that works with it, but it's hardly locked down to *that* specific hardware.
Mod me down with all of your hatred and your journey towards the dark side will be complete!
Ok, maybe I'm a Unix guy, but was this really something that changed the network? I know a lot of people have it installed, and run webservers, etc on it (usually because they are forced to or don't know any better), but if you want to put this on there it just seems like there are others that should be there like Solaris, Red Hat, Suze, FreeBSD just to name a few.
No Sigs!
I'll grant them most of these entries, but Apache was clearly not the first free web server. NCSA httpd was the first, and Apache is a derivative of that. The two coexisted for a few years, during which period it was possible to switch between them without even changing the config file. I think NCSA httpd project finally expired around 1996.
OK, that was the good things. Now let's remember the bad things and how they started.
- Adware. Ah.... the Gator download manager (TM). Didn't you love this thing? It was free! Only it began displaying some ads in your computer. What could possibly go wrong?
- SPAM. Funny, the other day i began receiving mails about mortgage rates. Idiots, I'm too young for that. I'll ignore it, they're 1 in a 100.
- Popups. OK, this is getting annoying. I'll have to block images from these free websites like XOOM, Geocities, Angelfire and so on.
- Web viruses. The other day something weird went on. I went to a porn website, and the next day my PC began opening popups. WTF?
- Email viruses. Ack! All I did was open my mail on Outlook express!
It's funny. We take these things for granted, but I remember the days when they didn't exist AT ALL. It was a wonderful era. Also worthy of notice is that all of them (except popups) were possible thanks to Microsoft Windows(TM).
I agree with you on something. Remove skype and add MSN/YIM/AIM/ICQ.
These non-anonymous chat services changed the way we relate to people on the web, replacing the untrusty anonymous IRC. It gave the ability to chat to every joe user.
As mentioned in the article, Novell was doing hierachal directories long before (and better than) Win2k. LDAP in and of itself wouldn't count because it wasn't use to centralize network management like NDS and AD were. Even today, generic LDAP based network management pales in comparison to NDS (now eDirectory).
-matthew
"THERE IS NO JUSTICE, THERE IS ONLY ME." -Death
Yes, it is true. Mike Tyson could probably kick
Muhammad Ali's ass. Of course, Mike Tyson is also
nearly 20 years younger. So, who is the better boxer?
For as much email has been run through sendmail in
the last couple decades, I'm always disappointed at how
little respect it receives.
I built my first mail server in 1993 using sendmail.
It brought internet email to my company over a serial
uucp link. By 1996, sendmail was moving nearly 87,000
internet messages a day for our company (not bad for a
486DX4-100 with a whopping 32M RAM (64M?)).
Saying the latest mail software (qmail, postfix, etc.)
is better than something written in 1972 - 27 years ago -
isn't saying much. (Well, maybe: Duh!)
Heck, 27 computing years is like 350 human years.
So, before you complain about security holes (one
in the last two years?) or complexity (like any other
programming language, practice makes perfect), why don't
you tell me which mail transport software you used in
1975, 1985 or 1995. Then, follow that up with which
transports you expect to see a lot of in 2010 and 2020.
Matt
This article, like most articles of its type, contains misleading generalities and outright factual errors.
1) Apache was NOT the first free web server. Both CERN httpd and NCSA's httpd predate it, and both were free.
2) Netscape and Spyglass's version of Mosaic were the first commercial WEB BROWSERS. The article states that both were the first commercial GUI's. Last time I checked the first commercial GUI was to be found on the Xerox Star circa 1981. Terminology matters, when you do not use a term correctly you create confusion and/or make yourself look like an arse.
The problem with these sorts of articles, and the magazines in which they appear, is that they're being written by journalists. I can't tell you the number of times over the years that I've had the misfortune of reading something computer related in a magazine or newspaper and discovered multiple serious factual errors. I've come to accept this from periodicals that don't normally deal with computers or technology, but I'm pretty much fed up with finding errors in PC magazine on a regular ongoing basis.
Who are the people who write these articles? There are some people who are interested in computers but aren't quite there yet in terms of their understanding. Many are not blessed with "the knack" (http://home.pcisys.net/~tbc/sounds/dilknack.wav) Others are so blessed, but are still neophytes. Either way they're very good at creating and passing on erroneous information about computers and technology.
Lee
Muslim community leaders warn of backlash from tomorrow morning's terrorist attack.
Sendmail is one of the most successful remote-access programs ever.
Sendmail has provided the essential r00t access for hax0rz to improve their skills in the past. Before Linux was cheap and available, one had to go out, and like a predator, acquire one's operating system privs. Sendmail was teh great enabler. Though I have moved on to better and brighter things, I thank Alman, and Vixie, for their great success in bringing r00t to the large number of adolescents everywhere.
This is a fallacy, and one that Linus himself debunks in his auto-biography.
A monolithic program may look more complex and harder to maintain and secure (and I'll admit, I hate sendmail), having a HUNDRED binaries as part of this program would add an order of magnitude of complexity that is entirely unnecesary.
Think: While it is true that a singular, small program which does one task is simpler than a monolithic giant, the program (as a whole, encompassing all the small parts) will still need to do all the same stuff a monolithic program has to do, except now it has to deal with message passing between small binary executables, queueing or drop files, and a number of other issues where security is a concern.
It's not as simple as taking parts out of the whole design and implementing them independantly; adding "parts" to the "whole" creates issues which do not exist in the monolithic.
qmail is able to do this fairly well, but it only has about 4 or 5 executables, IIRC, and it is compiled very carefully against bernsteins' special stdio and other library files that he's hardened.
See also: Linux Kernel vs. Hurd or Minix.
~Will
sig?
Seriously, anyone who calls themselves an Old-Timer in a field that is barely over 60 years old, is either a former co-worker of Turing or Von Neumann - the only generation with any business adding the word "old" - or they don't have enough understanding of the field to qualify.
Operating Systems in general are relatively new things. MULTICS is "historic", but only in the sense that it isn't in use. It has many ideas I consider valuable today, and I wish it was easier to get hold of MULTICS code, but it is far from ancient.
The odds are fairly high, though, that most "old-timers" on Slashdot are from the Unix or even the CP/M generation. Some might even call themselves "old-timers" when they only really started with DOS 3.1 or even something as modern as Windows 3.0!
I predate CP/M - not by much - but that doesn't matter because I don't claim to be an Old-Timer. Experienced, sure. Aware, certainly. Old-timer? No. I can tell you what I saw - from the control center at Jodrel Bank's Lovell Telescope to Imperial Computer's minis at Daresbury, from dusty Forth manuals to robotics and micromice - the word was Small. Small was good. Small was in. Small made Smartware one of the best damn integrated packages of that era in computing - and it outperformed many later generation systems. Small made Acornsoft's "Elite" the hottest game ever published by any title, as a percentage of the userbase it sold to.
Not sure if PETSpeed was small & unit-based. Wouldn't surprise me. You couldn't fit much even in a 32K machine, so modules would be logical.
As for Linus -- we're talking about Torvalds, right? The one who produced Linux, probably the most modular (and therefore smallest) OS ever released on this planet? The one who gave up on monolithic maintenance because he couldn't scale, so modularized even the maintenance process? You'd use him as an illustration for monolithic design, given that he hasn't used that in Linux in God-lost-count number of years?
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)