Home » IMAP » Page 2


I think I’m going to take a couple of days off again…

There’s an old Oasis lyric that I’ve hated for most of the past four years, mostly because of the memories it conjurs up, and now it really bugs me that it seems appropriate: “‘Cos I need more time just to make things right.”
The other fragments of the song that have survived my efforts to blot them out also seem fitting. “Don’t go away… Say that you’ll stay… Forever and a day…”

So, to ward off those perpetual rumors/fears/whatever that I’m hanging things up, I’ll just say this. I’ve “hung it up for good” before. The longest it ever lasted was 6 months, and that time was due to serious injury. Right around that time, someone insinuated that I should hang it up. We haven’t spoken since that time, and I’m better for it and I don’t give a rip about how he feels.

I fully expect this break to last through Saturday, then run out of gas sometime late Sunday afternoon.

I’ve overextended myself the past month or so. I’m tired. My Web server is running fabulously (it never hiccups, so long as Union Electric keeps the power flowing) but I haven’t come up with an effective way to upload content to it or add new features. I can live with that.

Meanwhile, my mail server’s a royal piece of… Nah. That doesn’t go far enough. My mail server is a Backstreet Boys Fan. It runs like a 16 MHz 386, and I can’t tell if it’s a configuration problem or if it’s just overwhelmed with spam. No matter. I’m overwhelmed with spam. On a good day I get 7. On a bad day I get 60+. I got 38 copies of the same spam message from some stupid online casino Tuesday. I absolutely have to get some spam filters in place, and some priority filters in place.

So the mail needs to be archived, a bare-essentials mail server built (Linux 2.4.8 kernel, sendmail, IMAP, fetchmail, procmail, and whatever else those five things force me to install so they can run, all built from the newest sources of course, using the most aggressive compiler settings known to man), then the archives restored, then spam filters put in place and run. Then I will have regained my ability to communicate and will be able to do something about my guilt over having week-old e-mail sitting around unanswered because it’s buried in worthless spam.

I need to tend to my servers. I need to rebuild a couple of workstations. I really ought to try to salvage the Baseball Mogul season that’s sitting on the corrupted hard drive in one of those workstations… (Though I hesitate to call anything that runs Microsoft Wintendo 2000 a “workstation…”)

Meanwhile, a couple of other projects need to get done, and I just realized today that I haven’t talked to Gatermann in more than a week and for all that group of friends knows, I’ve run off to the ends of the Earth only to find an Internet cafe, so I continued posting. I need to do something about that too.

I’ll be back. I have the same love/hate relationship with writing that most writers have. It’s like breathing after running a couple of miles on a brisk day in early March. I always hated breathing after that, because it hurt so badly. But no matter how much it hurt, I couldn’t stop.

Setting up Freesco for port forwarding

It’s a little late, but here’s how Gatermann and I got a Web server running behind a Freesco-based router. Freesco, despite the name, is a micro-distribution of Linux (based on the 2.0.x kernel) that offers firewalling, NAT, caching DNS, port forwarding, a lightweight Web server, and print services on a single floppy. Requirements are minimal; it’d run on a 386 with 8 megs of RAM, a floppy drive, and a pair of NE2000 NICs. For performance and ease of setup, I recommend a P75 (or faster, but a P75’s overkill; the main reason to use it is to get PCI) with a pair of PCI NICs and 8 megs.
What NICs do I recommend? Avoid the new Netgear FA311. The older FA310 worked fabulously, but Freesco doesn’t provide a module for the FA311’s NatSemi chipset, at least not yet. (The source code for a module is available at scyld.com and it’s compatible with the 2.0.x kernel, but compiling a kernel module isn’t a trivial operation for most of Freesco’s audience.) I’d probably go with a Realtek 8139-based card like a D-Link DFE-530TX+, a recent 3Com PCI card, or a PCI NE2000 clone. There’s a modules archive you can download that supports most other common NICs. A pair of D-Links, a P75 board, a floppy and this disk ought to give you nearly plug-and-play operation.

Enough of that. Here are the answers to the questions Freesco asked, in order.

Boot off the floppy. When it asks what you want to make with it, select ethernet router. Hostname doesn’t matter. Accept default for domain name, unless you’ve registered a domain for your LAN.

Don’t detect modems. Select two network cards. If you are using PCI cards, answer 0 to next four questions (IRQ, I/O). If you’re using ISA cards, enter the addresses and IRQs the cards use. DHCP? Depends on your ISP.

The first card’s name is eth0. (This is the card for your cable/dsl modem). Don’t use dhcp logging. Don’t update DNS by DHCP. Second card is eth1. Give it an IP address (10.x.x.x is fine, which is Freesco’s default; normally I use 192.168.1.x network and put my router on Network mask will almost always be I don’t configure for DHCP, so I don’t give it an IP range. if you want one, tell it the range of addresses you want to reserve. The fewer the better, for memory purposes, especially if you’ve only got 8 MB of RAM in the box.

Caching DNS? Answer S (secure). Don’t log.

Enable DHCP? Depends. If you don’t want to configure your LAN manually, DHCP is nice. If your LAN is already configured, DHCP is probably more trouble than it’s worth.

Public HTTP server. Answer Y. Default is S. Port 80. (You might be able to get away with answering N here, and you’ll save a little memory. DO NOT answer S–you’ll never forward port 80 if you do.)

Time server via HTTP? No.

Print server. No.

Telnet server. no.

Screensaver/spindown? 5 min is fine.

Swap file–0 if you have 8 mb or more. I suppose you could run Freesco on some tiny machines if you put in a small hard drive and enabled the swap file, but as cheap as a P75 with a pair of 4-meg SIMMs is these days, I wouldn’t bother.

Extra modules/programs? No.

Log: take defaults.

Host gateway–depends on ISP. Check one of your other PCs and use it.

Primary/secondary DNS. Use your ISP’s. Proxy, probably none. Check your ISP.

Export services? YES. This is the magic forwarding formula.

Now, assuming your web server is on, you’d use this line in config:


If you want to export other services, like, say, IMAP on port 143, add additional lines, subbing in the appropriate port and IP address. (HTTP is port 80.)

Pick a root password and web admin password, save configuration and reboot. You’re up and going.

Now, to configure your Windows boxes to get their Internet connection through your lovely what-was-old-is-new-again Freesco router, just open your TCP/IP settings, give it an IP address on the same subnet as your Freesco router if it doesn’t already have one, and set your gateway and DNS to the address you gave your Freesco router.

Voila. Configure your system’s BIOS for keyboardless operation if it has such an option, then take the keyboard and monitor away, write-protect the floppy and make a backup of it just in case (or burn it to a bootable CD if the machine is capable of booting off CD and you have an old drive to put in it) stick the box in a corner somewhere, and forget about it. If you have a power failure, it’ll reboot and happily start itself up again. As for stability, I find Freesco, in combination with decent hardware, is more stable than the hardware routers that are popular these days. Since it has a caching DNS, it’ll usually give you better performance too. And since you can probably build one with parts you have laying around, it’s cheaper.

Port forwarding with Linux

It’s Tuesday. I can’t wait for the weekend. Hey, at least this week we get a little break on Wednesday, at least in the States.
I posted some mail last night. Among those was a request that I reveal some of my Linux server-at-home secrets. I think I’ve sufficiently covered the creation of mail and Web servers, but I’ll go back and look some other time, when my brain’s less fried. I spent the day trying to make bootable Linux CDs. I’m thankful for CD-RWs, because I would have toasted about 10 CD-Rs in that process. I’ve found a Web site at work that talks all about it; I’ll refrain from calling it great until I figure out whether all of its steps actually work. I have made one successful bootable CD using the process, but it wouldn’t do everything I wanted. When I subbed in my own kernel that could do everything I wanted and left things like amateur radio support behind (just what I always wanted… a HAM-enabled Linux boot CD. Be still, my heart!) I got various different error messages. So not only am I wrong, I’m inconsistently wrong.

Anyway, let’s talk about firewalling. I don’t write firewalling scripts by hand; I let an expert do it. Then I go in and make slight modifications. My favorite method by far is to use PMFirewall, which asks you a bunch of nice questions and then writes a script. At present it only works with 2.2-based distros (a version for 2.4 is in alpha). If you want to do some forwarding, all you have to do is edit rc.firewall and add a couple of lines (this example assumes you’re running a Web server on, port 80):

echo "1" > /proc/sys/net/ipv4/ip_forward #enable IP forwarding
/usr/sbin/ipmasqadm portfw -a -P tcp -L $IPADDR 80 -R 80 #forward Web services to port 80 on

If you’re also running IMAP services on the same box, you can theoretically open it up with this line (I haven’t tried anything like this yet):

/usr/sbin/ipmasqadm portfw -a -P tcp -L $IPADDR 143 -R 143 #forward IMAP to port 143 on

Forwarding with Freesco is supposed to be easy but I’ve never actually done it yet. I’ll have to play around with it, on someone else’s cable or DSL connection of course (we wouldn’t want to keep anyone from reading these pages, after all). I believe Freesco is still 2.0-based, and firewalling and forwarding has changed with each major kernel revision since 2.0. It may have changed some before that too, for all I know, but back in those days I was fighting Slackware on 486s and deciding I hated Linux. It wasn’t until 1997 when a coworker gave me a copy of Red Hat 5.2 that I changed my mind and realized I didn’t hate Linux, I hated Slackware.

E-scape from the Hotel California…

Escaping Microsoft’s Hotel California. For lack of any other available alternative, I started using Outlook Express for mail about 18 months ago. It’s a decent mail client, does most of what I want–I don’t want much–and doesn’t do too terribly many things I don’t want it to. But it’s Microsoft. It runs on Windows. Its file formats are proprietary. It forces me to read my mail with the same workstation all the time. Migration makes me leave the mail behind. Most of it I want to leave behind, but do I want to sort it? NO! OK then. What to do?
Make an IMAP-enabled mail server out of a deprecated old PC and move all that mail over to it, that’s what. I tried to do this with TurboLinux but none of my mail clients wanted to talk to it. Since all of the books I have talk about Red Hat, I went with it, and it worked.

Here’s what I did. Install basic Red Hat. Include sendmail, procmail, fetchmail, imap. I pulled out all the XFree86 stuff. GUIs are for workstations. Command lines are for servers (and for workstations where you expect to get any work done quickly). Actually, I also pulled out just about everything else it would allow. A secure installation is a minimalist installation. After installation, edit /etc/inetd.conf. Uncomment imap line, save and exit. (I like pico, but you can do it with vi if that’s all you’ve got–find the line, delete the comment character, then save by hitting ZZ.) Bounce inetd with /etc/rc.d/rc3.d/inet stop ; /etc/rc.d/rc3.d/inet start. Create a user account with adduser [name] ; passwd [password].

Connect to your new IMAP server. For now, just use your ISP’s existing mail server for outgoing mail; use your IMAP server for incoming. Your username and password are the name/password you just created. After a brief delay, you should see your empty inbox, and you can start dragging stuff to it.

It went great for me. I created a new IMAP folder, opened one of OE’s folders, dragged all the contents over to the IMAP folder, and bingo! They moved. Read status and date were preserved too. (I’ve seen IMAP servers that wouldn’t do that.) I switched to another PC that had OE loaded and connected to my new mail server via IMAP and read some messages. Fantabulous.

Theoretically, I can go to my DSL router and forward port 143 to my mail server and read my mail from the outside.

Now, if you want to actually use your mail server to send mail, that gets trickier–you’ve gotta configure sendmail for that. The out-of-box setup is too secure to just use. Open /etc/mail/access and add your LAN to it, like so:

172.16.5 RELAY

Of greater interest is the fetchmail/procmail combo. You can use fetchmail to automatically go grab mail from the 47 mail accounts you have, then use procmail to sort it and filter out some spam.

To configure fetchmail, create the file /root/.fetchmailrc and chmod it to 0600. Here’s a very basic configuration:

poll mailserver.myisp.com
with protocol pop3
username myname password mypassword is my_name_on_my_linux_box

And finally, what’s the point of running your own mail server if you don’t spam filter it? There are lots of ways to go about it. I’m experimenting with this method. It uses procmail, which is called by sendmail, which is called by fetchmail. See how all this works?

If you want to get really smooth, you can even block mail before you download it with a program called Mailfilter. You probably don’t want to get as fancy with Mailfilter as people do with procmail, but you can use Mailfilter to search for certain key words or phrases like (checking my spam folder) viagra, mortgage, “fire your boss,” “lose weight” and delete them before you waste time and bandwidth downloading them. I’ve read estimates that spam traffic costs ISPs an average of $3 per month per user. Mailfilter won’t save your ISP very much, since the mail’s already been routed through its network and is just on its very last leg of the trip, but it’ll save them a little, and it’ll save you some bandwidth and time, so it’s probably worth it.

So if you’re looking to leave Outlook and/or Outlook Express all behind, or at least give yourself the option to use a different client, here’s the way out. It’s not too terribly difficult. And you gain an awful lot in the process: mail in a standardized, open format; redundancy; ease and versatility of backup (just schedule a cron job that tars it up and does stuff with it); the ability to very, very quickly search all of your mail with the Unix grep command (just log in, type grep -r [search string] * | more, and find what you’re looking for instantly) and far, far better mail filtering options.

And it’s infinitely cheaper (and more secure) than Exchange.


Troubleshooting Windows NT and Outlook. This one had me totally stumped. We have a PC that’s always been a little flaky, but very recently Outlook just stopped working reliably. Access violations every 5-20 minutes, always at the same address, became the norm. In addition, it would claim to run out of memory every time you tried to read anything other than a plain-text message.

I traced the latter problem to a corrupt normal.dot file; he was using Word as his e-mail editor and as his format for fancy messages.

Finally, I coaxed another error message out of it while logged in with an administrative account: Unable to find or register vbscript.dll. Fine. Time to uninstall, run Eraser 97 to get rid of the last of those bits, then reinstall both Office 97 and Outlook 98.

Vbscript.dll still didn’t register, so I grabbed a copy off a working PC, dropped it in WinntSystem32, and issued the command regsvr32 vbscript.dll to register it. It took.I launched Outlook 98, Word 97, and Internet Explorer 5, since that combination seems best able to coax whatever errors out of Outlook I’m going to get. I configured Outlook to go read my mail account via IMAP and waited. Access violation once again.

So I loaded a program we have from Seagate Software, called Modules.exe I believe. It tells you exactly what’s loaded in memory and at what address. Being able to count in hexadecimal doesn’t get you any dates, but it sometimes helps you troubleshoot a problem. Plus it lets you make jokes about how much more fun it makes you at parties. So I sort by memory address, look at the ranges, and find the culprit. I forget the name of the DLL offhand, but the description Modules gave was “Outlook Network Folders.” Net Folders. Net Folders: one of the best but most poorly implemented ideas Microsoft ever foisted upon the unsuspecting masses.

So I copied that DLL over from the properly-working machine, then used the command-line utility fc.exe to compare them–if they’re different, I’ve found my culprit. I compare them, and find they’re identical.

That leaves two possibilities: Either the OS is corrupt beyond repair and just needs a clean re-install from scratch, or we’ve got a hardware problem. I find it suspicious that the problem always occurs at the same address, so I take the address, translate it into decimal using Windows’ calculator, then divide it by 1024 and then by 1024 again to get the address in megabytes rather than in bytes, and I get… 1221-something. Ouch. In a machine with 64 MB of RAM? So I went and found a programmer to confirm whether I was running through the right mathematical process. Indeed I was. He asked what kind of memory translation Intel PCs and Windows NT do (he’s a VMS programmer).   Then I remembered NT uses a 4-gigabyte address space, regardless of how much physical memory is there. So much for getting a number between 0 and 64 that I could use to determine which DIMM in the system was bad.

So instead, I just swapped the DIMMs. The problem went away. A smoking gun! Problem is, which of the two was bad? I can’t just leave it that way, because eventually something else will break.

Popping out RAM Stress Test , from the unfortunately-named Ultra-X, Inc., is the best way I’ve found to conclusively find a bad module. So I ran it for a couple of hours. Nothing. So I left it to run over the weekend. Hopefully that will turn up the problem by the time I return to the office on Monday.

And of course the PC in question is the one on the desk of one of our most important executives. Figures.

A minor change. Seeing as I have a growing number of British readers, still cursed with dialup connections, and since I tend to write really long, I switched to a three-day view rather than the seven-day view I’ve been using. Older pieces are of course accessible through the calendar and through the Best of this Page link, both to the left.

Mac mice, PC data recovery

A two-button Mac mouse!? Frank McPherson asked what I would think of the multibutton/scroll wheel support in Mac OS X. Third-party multibutton mice have been supported via extensions for several years, but not officially from Ye Olde Apple. So what do I think? About stinkin’ time!

I use 3-button mice on my Windows boxes. The middle button double-clicks. Cuts down on clicks. I like it. On Unix, where the middle button brings up menus, I’d prefer a fourth button for double-clicking. Scroll wheels I don’t care about. The page up/down keys have performed that function just fine for 20 years. But some people like them; no harm done.

Data recovery. One of my users had a disk yesterday that wouldn’t read. Scandisk wouldn’t fix it. Norton Utilities 2000 wouldn’t fix it. I called in Norton Utilities 8. Its disktool.exe includes an option to revive a disk, essentially by doing a low-level format in place (presumably it reads the data, formats the cylinder, then writes the data back). That did the trick wonderfully. Run Disktool, then run NDD, then copy the contents to a fresh disk immediately.

So, if you ever run across an old DOS version of the Norton Utilities (version 7 or 8 certainly; earlier versions may be useful too), keep them! It’s something you’ll maybe need once a year. But when you need them, you need them badly. (Or someone you support does, since those in the know never rely on floppies for long-term data storage.) Recent versions of Norton Utilities for Win32 don’t include all of the old command-line utilities.

Hey, who was the genius who decided it was a good idea to cut, copy and paste files from the desktop? One of the nicest people in the world slipped up today copying a file. She hit cut instead of copy, then when she went to paste the file to the destination, she got an error message. Bye-bye file. Cut/copy-paste works fine for small files, but this was a 30-meg PowerPoint presentation. My colleague who supports her department couldn’t get the file back. I ride in on my white horse, Norton Utilities 4.0 for Windows in hand, and run Unerase off the CD. I get the file back, or so it appears. The undeleted copy won’t open. On a hunch, I hit paste. Another copy comes up. PowerPoint chokes on it too.

I tried everything. I ran PC Magazine’s Unfrag on it, which sometimes fixes problematic Office documents. No dice. I downloaded a PowerPoint recovery program. The document crashed the program. Thanks guys. Robyn never did you any harm. Now she’s out a presentation. Not that Microsoft cares, seeing as they already have the money.

I walked away wondering what would have happened if Amiga had won…

And there’s more to life than computers. There’s songwriting. After services tonight, the music director, John Scheusner, walks up and points at me. “Don’t go anywhere.” His girlfriend, Jennifer, in earshot, asks what we’re plotting. “I’m gonna play Dave the song that he wrote. You’re more than welcome to join us.”

Actually, it’s the song John and I wrote. I wrote some lyrics. John rearranged them a little (the way I wrote it, the song was too fast–imagine that, something too fast from someone used to writing punk rock) and wrote music.

I wrote the song hearing it sung like The Cars, (along the lines of “Magic,” if you’re familiar with their work) but what John wrote and played sounded more like Joe Jackson. Jazzy. I thought it was great. Jennfier thought it was really great.

Then John tells me they’re playing it Sunday. They’re what!? That will be WEIRD. And after the service will be weird too, seeing as everybody knows me and nobody’s ever seen me take a lick of interest in worship music before.

I like it now, but the lyrics are nothing special, so I don’t know if I’ll like it in six months. We’ll see. Some people will think it’s the greatest thing there ever was, just because two people they know wrote it. Others will call it a crappy worship song, but hopefully they’ll give us a little credit: At least we’re producing our own crappy worship songs instead of playing someone else’s.

Then John turns to me on the way out. “Hey, you’re a writer. How do we go about copyrighting this thing?” Besides writing “Copyright 2000 by John Scheusner and Dave Farquhar” on every copy, there’s this.  That’s what the Web is for, friends.


Note: I post this letter without comment, since it’s a response to a letter I wrote. My stuff is in italics. I’m not sure I totally agree with all of it, but it certainly made me think a lot and I can’t fault the logic.

From: John Klos
Subject: Re: Your letter on Jerry Pournelle’s site

Hello, Dave,

I found both your writeup and this letter interesting. Especially interesting is both your reaction and Jerry’s reaction to my initial letter, which had little to do with my server.To restate my feelings, I was disturbed about Jerry’s column because it sounded so damned unscientific, and I felt that he had a responsibility to do better.
His conclusion sounded like something a salesperson would say, and in fact did sound like things I have heard from salespeople and self-promoted, wannabe geeks. I’ve heard all sorts of tales from people like this, such as the fact that computers get slower with age because the ram wears out…

Mentioning my Amiga was simply meant to point out that not only was I talking about something that bothered me, but I am running systems that “conventional wisdom” would say are underpowered. However, based upon what both you and Jerry have replied, I suppose I should’ve explained more about my Amiga.

I have about 50 users on erika (named after a dear friend). At any one moment, there are anywhere from half a dozen to a dozen people logged on. Now, I don’t claim to know what a Microsoft Terminal Server is, nor what it does, but it sounds something like an ’80s way of Microsoft subverting telnet.

My users actually telnet (technically, they all use ssh; telnet is off), they actually do tons of work is a shell, actually use pine for email and links (a lynx successor) for browsing. I have a number of developers who do most of their development work in any of a number of languages on erika (Perl, C, C++, PHP, Python, even Fortran!).

Most of my users can be separated into two groups: geeks and novices. Novices usually want simple email or want to host their domain with a minimum of fuss; most of them actually welcome the simplicity, speed, and consistency of pine as compared to slow and buggy webmail. Who has used webmail and never typed a long letter only to have an error destroy the entire thing?

The geeks are why sixgirls.org got started. We all
had a need for a place
to call home, as we all have experienced the nomadic life of being a geek
on the Internet with no server of our own. We drifted from ISP to ISP
looking for a place where our Unix was nice, where our sysadmins listened,
and where corporate interests weren’t going to yank stuff out from underneath us at any moment. Over the years, many ISPs have stopped
offering shell access and generally have gotten too big for the comfort of

If Jerry were replying to this now, I could see him saying that shells are
old school and that erika is perhaps not much more than a home for  orphans and die-hard Unix fans. I used to think so, too, but the more novice users I add, the more convinced I am that people who have had no shell experience at all prefer the ease, speed, and consistency of the shell
over a web browser type interface. They’re amazed at the speed. They’re
surprised over the ability to instantly interact with others using talk and ytalk.

The point is that this is neither a stopgap nor a dead end; this IS the
future. I read your message to Jerry and it got me thinking a lot. An awful
lot. First on the wisdom of using something other than what Intel calls a server, then on the wisdom of using something other than a Wintel box as a server. I probably wouldn’t shout it from the mountaintops if I were doing it, but I’ve done it myself. As an Amiga veteran (I once published an article in Amazing Computing), I smiled when I saw what you were doing with your A4000. And some people no doubt are very interested in that. I wrote some about that on my Weblogs site (address below if you’re interested).

I am a Unix Systems Administrator, and I’ve set up lots of servers. I made
my decision to run everything on my Amiga based upon several
One, x86 hardware is low quality. I stress test all of the servers I
build, and most x86 hardware is flawed in one way or another. Even if
those flaws are so insignificant that they never affect the running of a
server, I cannot help but wonder why my stress testing code will run just
fine on one computer for months and will run fine on another computer for
a week, but then dump a core or stop with an error. But this is quite
commonplace with x86 hardware.

For example, my girlfriend’s IBM brand FreeBSD computer can run the stress testing software indefinitely while she is running the GIMP, Netscape, and all sorts of other things. This is one of the few PCs that never has any problems with this stress testing software. But most of the other servers I set up, from PIIIs, dual processor PIIIs and dual Celerons, to Cyrix 6×86 and MII, end up having a problem with my software after anywhere from a few days to a few weeks. But they all have remarkable uptimes, and none crash for any reason other than human error (like kicking the cord).

However, my Amigas and my PowerMacs can run this software indefinitely.

So although I work with x86 extensively, it’s not my ideal choice. So what
else is there? There’s SPARC, MIPS, m68k, PowerPC, Alpha, StrongARM… pleanty of choices.

I have a few PowerMacs and a dual processor Amiga (68060 and 200 mhz PPC 604e); however, NetBSD for PowerMacs is not yet as mature as I need it to be. For one, there is no port of MIT pthreads, which is required for MySQL. Several of my users depend on MySQL, so until that is fixed, I can’t consider using my PowerMac. Also, because of the need to boot using Open Firmware, I cannot set up my PowerMac to boot unattended. Since my machine is colocated, I would have to be able to run down to the colocation facility if anything ever happened to it. That’s
fine if I’m in the city, but what happens when I’m travelling in Europe?

SPARC is nice, but expensive. If I could afford a nice UltraSPARC, I
would. However, this porject started as a way to have a home for
geeks; coming up with a minimum of $3000 for something I didn’t even plan to charge for wasn’t an option.

Alpha seems too much like PC hardware, but I’d certainly be willing to
give it a try should send me an old Alpha box.

With MIPS, again, the issue is price. I’ve always respected the quality of
SGI hardware, so I’d definitely set one up if one were donated.

StrongARM is decent. I even researched this a bit; I can get an ATX
motherboard from the UK with a 233 mhz StrongARM for about 310 quid. Not too bad.

But short of all of that, I had a nice Amiga 4000 with a 66 mhz 68060, 64
bit ram, and wide ultra SCSI on board. Now what impresses me about this
hardware is that I’ve run it constantly. When I went to New Orleans last
year during the summer, I left it in the apartment, running, while the
temperatures were up around 100 degrees. When I came back, it was
fine. Not a complaint.

That’s the way it’s always been with all of my Amigas. I plug them in,
they run; when I’m done, I turn off the monitor. So when I was considering
what computer to use as a server when I’d be paying for a burstable 10
Mbps colocation, I wanted something that would be stable and consistent.

 Hence Amiga.

One of my users, after reading your letter (and, I guess, Jerry’s),
thought that I should mention the load average of the server; I assume
this is because of the indirectly stated assumption that a 66 mhz 68060 is
just squeaking by. To clarify that, a 66 mhz 68060 is faster per mhz than
any Pentium by a measurable margin when using either optimised code (such as a distributed.net client) or straight compiled code (such as LAME). We get about 25,000 hits a day, for a total of about 200 megs a day, which accounts for one e

ighth of one percent of the CPU time. We run as a Stratum 2 time server for several hundred computers, we run POP and IMAP services, sendmail, and we’re the primary nameserver for perhaps a hundred machines. With a distributed.net client running, our load average hovers arount 1.18, which means that without the dnet client, we’d be idle most of the time.

If that weren’t good enough, NetBSD 1.5 (we’re running 1.4.2) has a much
improved virtual memory system (UVM), improvements and speedups in the TCP stack (and complete IPv6 support), scheduler enhancements, good softdep support in the filesystem (as if two 10k rpm 18 gig IBM wide ultra drives aren’t fast enough), and more.

In other words, things are only going to get better.

The other question you raise (sort of) is why Linux gets so much more
attention than the BSD flavors. I’m still trying to figure that one
out. Part of it is probably due to the existance of Red Hat and
Caldera and others. FreeBSD gets some promotion from Walnut
Creek/BSDi, but one only has to look at the success of Slackware to
see how that compares.

It’s all hype; people love buzz words, and so a cycle begins: people talk
about Linux, companies spring up to provide Linux stuff, and people hear
more and talk more about Linux.

It’s not a bad thing; anything that moves the mainstream away from
Microsoft is good. However, the current trend in Linux is not good. Red
Hat (the company), arguably the biggest force in popularising Linux in the
US, is becoming less and less like Linux and more and more like a software company. They’re releasing unstable release after unstable release with no apologies. Something I said a little while ago, and someone has been using as his quote in his email:
In the Linux world, all of the major distributions have become
companies. How much revenue would Red Hat generate if their product was flawless? How much support would they sell?

I summarise this by saying that it is no longer in their best interest to
have the best product. It appears to be sufficient to have a working
product they can use to “ride the wave” of popularity of Linux.

I used Linux for a long time, but ultimately I was always frustrated with
the (sometimes significant) differences between the distributions, and
sometimes the differences between versions of the same distribution. Why
was it that an Amiga running AmigaDOS was more consistent with Apache and Samba docs than any particular Linux? Where was Linux sticking all of
these config files, and why wasn’t there documentation saying where the
stuff was and why?

When I first started using BSD, I fell in love with its consistency, its
no bull attitude towards ports and packa
ges, and its professional and
clean feel. Needless to say, I don’t do much linux anymore.

It may well be due to the people involved. Linus Torvalds is a
likeable guy, a smart guy, easily identifiable by a largely computer
illiterate press as an anti-Gates. And he looks the part. Bob Young is
loud and flambouyant. Caldera’s the company that sued Microsoft and probably would have won if it hadn’t settled out of court. Richard
Stallman torques a lot of people off, but he’s very good at getting
himself heard, and the GPL seems designed at least in part to attract
attention. The BSD license is more free than the GPL, but while
freedom is one of Stallman’s goals, clearly getting attention for his
movement is another, and in that regard Stallman succeeds much more than the BSD camp. The BSD license may be too free for its own good.

Yes, there aren’t many “figureheads” for BSD; most of the ones I know of
don’t complain about Linux, whereas Linux people often do complain about the BSD folks (the major complaint being the license).

I know Jerry pays more attention to Linux than the BSDs partly because Linux has a bigger audience, but he certainly knows more about Linux than about any other Unix. Very soon after he launched his website, a couple of Linux gurus (most notably Moshe Bar, himself now a Byte columnist) started corresponding with him regularly, and they’ve made Linux a reasonably comfortable place for him, answering his questions and getting him up and going.

So then it should be their responsibility, as Linux advocates, to give
Jerry a slightly more complete story, in my opinion.

As for the rest of the press, most of them pay attention to Linux only because of the aforementioned talking heads. I have a degree in journalism from supposedly the best journalism school in the free world, which gives me some insight into how the press works (or doesn’t, as is usually the case). There are computer journalists who get it, but a g

ood deal of them are writing about computers for no reason in particular, and their previous job and their next job are likely to be writing about something else. In journalism, if three sources corroborate something, you can treat it as fact. Microsoft-sympathetic sources are rampant, wherever you are. The journalist probably has a Mac sympathy since there’s a decent chance that’s what he uses. If he uses a Windows PC, he may or may not realize it. He’s probably heard of Unix, but his chances of having three local Unix-sympathetic sources to use consistently are fairly slim. His chances of having three Unix-sympathetic sources who agree enough for him to treat what they say as fact (especially if one of his Microsofties contradicts it) are probably even more slim.

Which furthers my previous point: Jerry’s Linux friends should be more
complete in their advocacy.

Media often seems to desire to cater to the lowest common denominator, but it is refreshing to see what happens when it doesn’t; I can’t stand US
news on TV, but I’ll willingly watch BBC news, and will often learn more
about US news than if I had watched a US news program.

But I think that part of the problem, which is compounded by the above, is
that there are too many journaists that are writing about computers,
rather than computer people writing about computers.

After all, which is more presumptuous: a journaist who thinks that he/she
can enter the technical world of computing and write authoritatively about
it, or a computer person who attempts to be a part time journalist? I’d
prefer the latter, even if it doesn’t include all of the accoutrements
that come from the writings of a real journalist.

And looking at the movement as a whole, keep in mind that journalists look for stories. Let’s face it: A college student from Finland writing an operating system and giving it away and millions of people thinking it’s better than Windows is a big story. And let’s face it, RMS running
around looking like John the Baptist extolling the virtues of something called Free Software is another really good story, though he’d get a lot more press if he’d talk more candidly about the rest of his life, since that might be the hook that gets the story. Can’t you see this one now?

Yes. Both of those stories would seem much more interesting than, “It’s
been over three years and counting since a remote hole was found in
OpenBSD”, because it’s not sensationalistic, nor is it interesting, nor
can someone explain how you might end up running OpenBSD on your
appliances (well, you might, but the fact that it’s secure means that it’d
be as boring as telling you why your bathtub hasn’t collapsed yet).

Richard Stallman used to keep a bed in his office at the MIT Artificial Intelligence Lab.

He slept there. He used the shower down the hall. He didn’t have a home outside the office. It would have distracted him from his cause: Giving away software.

Stallman founded the Free Software movement in 1983. Regarded by many as the prophet of his movement (and looking the part, thanks to his long, unkempt hair and beard), Stallman is both one of its most highly regarded programmers and perhaps its most outspoken activist, speaking at various functions around the world.

Linux was newsworthy, thanks to the people behind it, way back in 1993 when hardly anyone was using it. Back then, they were the story. Now, they can still be the story, depending on the writer’s approach.

If there are similar stories in the BSD camp, I’m not aware of them. (I can tell you the philosophical differences between OpenBSD,  NetBSD and FreeBSD and I know a little about the BSD directory structure, but that’s where my knowledge runs up against its limits. I’d say I’m more familiar with BSD than the average computer user but that’s not saying much.) But I can tell you my editor would have absolutely eaten this up. After he or she confirmed it wasn’t fiction.

The history is a little dry; the only “juicy” part is where Berkeley had
to deal with a lawsuit from AT&T (or Bell Labs; I’m not doing my research
here) before they could make their source free.

Nowadays, people are interested because a major layer of Mac OS X is BSD, and is taken from the FreeBSD and NetBSD source trees. Therefore, millions of people who otherwise know nothing about BSD or its history will end up running it when Mac OS X Final comes out in January; lots of people already are running Mac OS X Beta, but chances are good that the people who bought the Beta know about the fact that it’s running on BSD.

And it’s certainly arguable that BSD is much more powerful and robust than Windows 2000. So there’s a story for you. Does that answer any of your question?

Yes; I hope I’ve clarified my issues, too.

Neat site! I’ll have to keep up on it.

John Klos