photo-rbt.jpg (2942 bytes)

Email Robert

Daynotes Journal

Week of 21 June 1999

Sunday, June 27, 1999 09:32

A (mostly) daily journal of the trials, tribulations, and random observations of Robert Bruce Thompson, a writer of computer books.


 

 

 

Search this Site

TTG Home

Robert Home

Daynotes Home

Special Reports

Current Topics


Monday, June 21, 1999

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]


Yet another night sleeping on the floor. Barbara is painting the master bedroom and bath. The house has four bedrooms on the main level, but one is my office, one is Barbara's office, and the guest bedroom is covered up with stuff temporarily relocated to allow painting. So we're sleeping in the den. Barbara gets the sofa. Our younger Border Collie, Duncan, gets the love seat. The older Border Collie, Kerry, is sleeping downstairs in my mother's apartment. That leaves the floor for me. Fortunately, I have two large dog beds available, which actually turn out to be very comfortable. Of course, the first thing I do in the morning is sit up and scratch with my hind leg. I could live with that, but I found myself marking a fire hydrant the other night.

* * * * *

I finally started to play with CD-RW yesterday evening. I'd picked up some CD-RW disks Saturday evening. I wanted to use Adaptec DirectCD, which is the best packet-writing software out there, but the Smart & Friendly drive didn't come with a copy of it. But I had an older copy of DirectCD on diskette, so I dug it out and installed it on the test-bed system where the CD-RW drive is installed. This was DirectCD v 2.0, with a 1997 date on it, so I wasn't entirely surprised when the installation finished and it wouldn't recognize my drive.

I remembered seeing a downloadable update to version 2.5 on the Adaptec web site, so I went over and downloaded it. A 5 MB patch to a 2 MB program. Once the download finished, I realized that I was in one of those "you can't get there from here" situations. The update was stored on a network drive, and the test-bed system wasn't connected to the network.

When I installed the Intel InBusiness 8-Port 10/100 hub, I had to steal the Ethernet cable from the test-bed machine to connect the Intel hub to the old 3Com 10BaseT hub so that I could continue to access my Internet gateway machine, which connects to the network via 10Base2 coax on the 3Com hub. That left me one 100BaseT cable short. I know I have spare 100BaseT cables around here. I can even remember what color they are--bright yellow and hot pink. But I sure couldn't find any of them.

By then, it was time to leave the house. We drove over to pick up Barbara's parents and headed for her sister's house for Fathers' Day dinner. Just before we were ready to leave there, I thought about my friend Steve Tucker. His computer spares closet is better stocked than the supply room of some computer stores. I figured he'd have a spare 100BaseT cable, so I called him. Sure enough, he had one which he was kind enough to give me, refusing payment.

When we got home, I connected the test-bed system back onto the network and installed the DirectCD 2.5 patch, which recognized the Smart & Friendly CD SpeedWriter Plus drive properly. I put a CD-RW disk in the drive and fired up the Wizard, which informed me that formatting the CD-RW disk would require 50 to 90 minutes. An hour or so later, the disk had formatted to 553 MB. The 100 MB or so of lost capacity is occupied by the UDF file system that allows the CD-RW disk to behave like an enormous floppy disk.

This drive re-writes at 2X, or 300 KB/s. That translates to 18 MB/min, which means that filling this disk to capacity will require about half an hour. I guess the next step is to fill the disk up and then find out which of the CD-ROM drives in my other computers can read it without having UDF extensions installed on the computer. 

There are actually two separate issues there: first, CD-RW disks are much less reflective than standard CDs or even CD-R disks, so many older drives cannot physically read the data from CD-RW disks. Second, even if a CD-ROM drive can physically read the disk, it may not understand the logical UDF formatting.

* * * * *

This from Jan Swijsen [qjsw@oce.nl]: 

I'm surprised that you're able to do a direct write from a hard disk on the same ATA channel as the burner.

It looks like a timing issue. A hard disk typically has fast seek and access times compared to a CD drive. A slow hard drive will still be faster than a fast CD drive even though its throughput may be lower. What happens looks like this: the Writer puts a request for data on the IDE chain, this passes the request to the CPU which passes it to the source (via the IDE), which fetches it and sends it back to memory (DMA) it sends a signal to the CPU when finished, this sends the signal (with DMA address) to the writer, which fetches the data from RAM (via DMA) to top up its buffer.

Any delay in this chain of events may cause a buffer under run (=yet another coaster). The access time of a CD drive may just be the breaking factor. Or the fantastic multitasking of Win9x (ROGL) may cause problems as well. With a fast accessing hard disk and a dedicated (or properly multitasking) system you eliminate the two most problematic factors. 

Because SCSI multitasks (real), some of the signals can pass 'concurrently' so the CPU may be informed of the memory location the source is writing to before all the data has arrived there and the writer may start reading before the source has actually finished writing. That is (probably) why SCSI is less prone to under runs.

This is a hypothesis of course and I don't know how to actually check it. The lesson is that what works on one PC may not work on another. You really have to love these boxes :) .

Svenson

Perhaps so, but I don't think there's any question that putting a CD burner on the same ATA channel as its source device(s) is a pretty bad idea. I've also noticed that operations seem to be much more reliable under Windows NT than under Windows 98, which may be due to the multi-tasking issue you mention.

 


 

 

 

Search this Site

TTG Home

Robert Home

Daynotes Home

Special Reports

Current Topics


Tuesday, June 22, 1999

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]


Still hard at work on the books. I spent half the day yesterday writing and the other half on the phone with vendors. There's some very interesting stuff just over the horizon. A lot of it isn't yet even sampling to the press, but I should have eval units of some pretty intriguing stuff arriving over the next two or three months. 

One fascinating product is the Castlewood ORB, a direct competitor to the Iomega Jaz removable hard drive. Where the 2 GB Jaz sells for about $500 and uses $125 disk cartridges, the 2.2 GB ORB will sell for $150 to $175 and use $30 cartridges. The Jaz has never sold like the Zip, and the ORB may be a Jaz-killer. I should have an ORB in-house for testing within the next six to eight weeks. If you're considering buying a Jaz, you might want to wait.

Back to the writing grind...

* * * * *

This from Edmund C. Hack [echack@crl.com]: 

The item attributed to the state senator from Georgia was not written by him. It is titled the "Bill of No Rights" and was written by Lewis Napper. His web site is at http://www.bserver.com/bunker/index.html and has a number of entertaining Libertarian viewpoint essays.

Edmund Hack \ "It's like Titanic - without the water!"
echack@crl.com
\"US Plus: We own the idea of America." - Firesign Theatre

Thanks. Quite an interesting web site. Needless to say, that's a guy I could get along with.


 

 

 

Search this Site

TTG Home

Robert Home

Daynotes Home

Special Reports

Current Topics


Wednesday, June 23, 1999

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]


You've probably already noticed the new Search function. This all started rather circularly. Paul Robichaux, a fellow O'Reilly author, posted a message to a thread on the Computer Book Publishing mailing list about FrontPage problems. That message pointed to my site and the ongoing discussions here. Rafe Colburn [rafe.colburn@interpath.net], another author who participates on that list, sent me the following.

I noticed that you were looking for a search engine for your personal Web site to replace the one provided with the FrontPage extensions. For my home page (http://rc3.org), I use Thunderstone's free search engine. They spider your site and host the search engine, you just have to publish a front end to the search engine on your site.

I believe you can have it respider whenever you like, but they automatically walk your site once a week.

Here's the URL:

http://index.thunderstone.com/texis/indexsite/

The search engine is very powerful and fast, and of course, it's ridiculously simple to maintain. You can also add your own header and footer to the results pages.

--

Rafe Colburn | work: http://interpath.net
Interpath Communications | personal: http://rc3.org
rafe.colburn@interpath.net | phone: 919.654.2247 x246

Thanks a million. That solved my problem instantly. I'm already set up with Thunderstone, they've completely indexed my site, and I've embedded the code in my pages (although it won't show up until I publish tomorrow). ["today" RBT]

* * * * *

So I went to the ../indexsite/ URL that Rafe mentioned, and signed up my site to be indexed by Thunderstone. They send back an init key via email that you need to enter as a part of the signup process. I'm not sure why they're so determined to prevent anyone but the site owner from requesting that the site be indexed, but that accomplishes their goal. 

I was rather surprised to find that by the time I'd completed the sign-up process, they'd already retrieved every page in my site, indexed it, and made it searchable. They've apparently changed their policies in at least one respect: Rafe mentions that they automatically re-index your site every week, but Thunderstone now says they do it every two weeks. Either way, it's no problem. Stuff two weeks or newer really doesn't need to be indexed, and I can always manually request an immediate re-indexing any time if that seems desirable.

They provide the HTML code to display the search box. I did some minor edits and dropped it into this page. I've left the main search page active, simply because I have so many other pages that point to it. All that page does, however, is display the same search box that will now appear on the weekly journal pages.

While I was editing this page to add the search box, I also decided to move some stuff around a bit, which accounts for the slight change in appearance.

* * * * *

Yesterday was an exciting day. It started about lunch time as I was sitting here writing away. All of a sudden, both the UPSs under my desk (a TrippLite 675 and an APC Smart-UPS 1100) started going berserk. There wasn't any obvious power failure, but nearly everything in my office runs on a UPS, so I might not have noticed. I called downstairs to see if my mother had noticed any power glitches, and she said the TV had gone off momentarily just then.

I came back to my office, where the UPSs were still beeping and clicking away. From their indicator lights, I thought perhaps we were having a brown-out or something. I shut down all the computers and turned off the UPSs. As I was crawling under the desk to turn off the TrippLite, I noticed a strong ozone smell--that nasty odor of burning insulation and high voltage trying to get to places it's not supposed to be.

My first thought was that it was the TrippLite 675. Natural enough, since this unit is a replacement for one that literally threw sparks the first time I plugged it in. After disconnecting all the power, I started crawling around under the desk, sniffing the bouquet of the TrippLite, the APC, and all the computers. Because the smell is so pervasive, I couldn't determine for sure which component was failing. At this point, my prime suspect is an old Mag 15" monitor, although it sits on top of my desk. The smell seems somewhat stronger there than elsewhere, but not overwhelmingly so.

At this point, I've powered everything back up, and I *still* can't tell where the smell originated. It hasn't dissipated, but at least it's getting no worse. As Barbara says, "So, you're just sitting there waiting for something to catch on fire?" Yup. Nothing feels hot, and I can't track the smell by nose alone, so I don't know what else to do, short of shutting down my entire network. This morning, the smell is still present, but less pervasive. I left the old Mag 15" monitor off overnight. Given some of the weird stuff that happened with my resource server yesterday, it may be the power supply in it rather than that monitor. We'll see.

* * * * *

The coprophages at GTE Wireless have finally motivated me to do something about dropping their service. I'll be damned if I'll keep paying them for screwing up my bill. Last month, I wasted half an hour getting them to correct my bill from $45.89 to $44.41. The lady swore that it was fixed and that it would stay fixed. I specifically asked her if she was sure I wouldn't get billed the following month for an underpayment. She swore I would not. This month's bill just showed up, and guess what? They billed me for $45.90 (it goes up yet another cent) for this month, plus the $1.48 they claim I underpaid them by for last month, plus a one cent finance charge. Were I not a peaceful man, I'd track them down and have a heart-to-heart chat with them.

Enough of this. We don't really need two cell phones now that Barbara works at home. I'll drop their service immediately and find someone else to buy cellular service from. Not that that will probably make much difference. They're all obnoxious. They're just like a government agency, which they might as well be. Instead of keeping their damned hands off cellular, the government has gone in and granted monopolies (well, duopolies, really) to the highest bidders. If they'd just let the free market take care of things, we'd all have reliable cellular service from our choice of any number of providers, and it would be a lot cheaper than it is now. I hate the government, and I hate cell phone companies. Compared to either of them, Microsoft is a shining example of how things should be done.

* * * * *

And then I got this delightful little missive to inform me that the pair Networks mail servers couldn't find support@pair.com. I don't know what they expect me to do about it. pair Networks has the most pathetic mail handling of any service I've encountered. To this day, I get all return receipts and bounces addressed to barbara@ttgnet.com. Other messages addressed to her are properly forwarded to her POP account at BellSouth.net, but return receipts and bounces end up in the general delivery mailbox for ttgnet.com, where I eventually POP them. pair Network's tech support explains this as a necessary practice to avoid mail loops. That's pretty lame. No other service provider I've dealt with seems to have this problem. I think they either need new SMTP software or someone who understands SMTP better. At any rate, here's the message.

Hi. This is the qmail-send program at mx1.pair.com.

I'm afraid I wasn't able to deliver your message to the following addresses.

This is a permanent error; I've given up. Sorry it didn't work out.

<support@pair.com>:

This message is looping: it already has my Delivered-To line. (#5.4.6)

--- Below this line is a copy of the message.

Return-Path: <thompson@ttgnet.com>
Received: (qmail 18894 invoked from network); 22 Jun 1999 17:17:00 -0000
Received: from smx.pair.com (209.68.1.56)
by mx1.pair.com with SMTP; 22 Jun 1999 17:17:00 -0000
Received: (qmail 11052 invoked by uid 915); 22 Jun 1999 17:17:00 -0000
Delivered-To: abuse@smx.pair.com
Received: (qmail 6069 invoked from network); 22 Jun 1999 02:02:01 -0000
Received: from mx2.pair.com (209.68.1.62)
by smx.pair.com with SMTP; 22 Jun 1999 02:02:01 -0000
Received: (qmail 5712 invoked by uid 915); 22 Jun 1999 02:02:01 -0000
Delivered-To: abuse@pair.com
Received: (qmail 5709 invoked from network); 22 Jun 1999 02:02:01 -0000
Received: from smx.pair.com (209.68.1.56)
by mx2.pair.com with SMTP; 22 Jun 1999 02:02:01 -0000
Received: (qmail 6066 invoked by uid 901); 22 Jun 1999 02:02:00 -0000
Delivered-To: support@smx.pair.com
Received: (qmail 10417 invoked from network); 21 Jun 1999 13:11:28 -0000
Received: from mx1.pair.com (209.68.1.60)
by smx.pair.com with SMTP; 21 Jun 1999 13:11:28 -0000
Received: (qmail 16715 invoked by uid 901); 21 Jun 1999 13:11:28 -0000
Delivered-To: support@pair.com
Received: (qmail 16712 invoked from network); 21 Jun 1999 13:11:28 -0000
Received: from mail1.bna.bellsouth.net (205.152.80.13)
by mx1.pair.com with SMTP; 21 Jun 1999 13:11:28 -0000
Received: from kerby (host-209-214-60-136.int.bellsouth.net [209.214.60.136])
by mail1.bna.bellsouth.net (8.8.8-spamdog/8.8.5) with SMTP id JAA29653
for <support@pair.com>; Mon, 21 Jun 1999 09:11:19 -0400 (EDT)
Return-Receipt-To: "Robert Bruce Thompson" <thompson@ttgnet.com>
Reply-To: <thompson@ttgnet.com>
From: "Robert Bruce Thompson" <thompson@ttgnet.com>
To: <support@pair.com>
Subject: Please remove FrontPage Extensions from my account
Date: Mon, 21 Jun 1999 09:10:56 -0400
Message-ID: <000301bebbe7$8037c290$a16fa8c0@ttgnet.com>
MIME-Version: 1.0
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook CWS, Build 9.0.2416 (9.0.2910.0)
Importance: Normal
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300
Disposition-Notification-To: "Robert Bruce Thompson" <thompson@ttgnet.com>
Sender: abuse@pair.com

My web site has apparently grown to the size that causes the FrontPage extensions to stop working reliably.

Please remove the FrontPage extensions from my site, ttgnet.com (server 109), as soon as possible.

Thank you.

Robert Bruce Thompson

thompson@ttgnet.com
http://www.ttgnet.com

I forwarded this message to urgent@pair.com yesterday afternoon at 1:30, shortly after I'd received it. So far, nothing but the automated response that says they received it...

* * * * *

Mid-afternoon: Just got mail from support@pair.com, saying that they don't know why mail to them bounced and that they've taken the FP extensions off my site. Ironic, considering that the last two times I've published, I've gotten all the way to the final "Published to.." prompt--the first times that's happened in a long time. At any rate, I'll try publishing this using Barbara's system and FP98 first. If that works, I'll give it a shot with my system and FP2000.

 


 

 

 

Search this Site

TTG Home

Robert Home

Daynotes Home

Special Reports

Current Topics


Thursday, June 24, 1999

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]


Well, this sucks. I tried publishing my web updates from Barbara's system, running FP98. As soon as I clicked Publish, FP came back and told me that the FP Extensions weren't running on the web server, and that I'd have to publish via ftp with the FTP Publishing Wizard. So far, so good. I expected that. What I didn't expect was to have problems getting an ftp connection established. 

I'm not at all sure the problem lies with FrontPage. I think it has something to do with my ftp configuration (or lack thereof) for WinGate. Barbara's machine accesses the Internet via the WinGate proxy server. I've never been able to get my CuteFTP client running on kerby to access ftp sites via WinGate, either, so the problem is probably with how I have WinGate set up. I tried mucking about some with the proxy configuration on the client and the ftp service configuration on the WinGate server, but nothing seemed to help. The next step was to try publishing from kerby, using FP2K. That didn't work, either. It couldn't make the ftp connection through the proxy.

The next step was to check WinGate help. It recommended installing the WinGate Internet Client, and said that I could just run the executable that I'd used to install WinGate on the server. That executable would recognize that a WinGate server existed elsewhere on the network and automatically install the WIC on the client. The only trouble was, I couldn't find the original distribution file.

I hit the WinGate web site to download the WinGate executable all over again. I learned that the WIC wasn't included in the original WinGate 3.0 distribution anyway, and that WIC requires SP4 on NT client machines. Barbara's and my systems are still running SP3. I was kind of holding out for SP5 to get out of beta, but I figured installing SP4 made sense, so I tried to download the WinGate executable. After entering the required name, email address, etc., I finally got to the download screen. There were several alternate sites listed for the download, but all of them returned a 404 error. Not good.

I had one remaining alternative. I can access the Internet just fine on the WinGate server itself, because access doesn't go through WinGate. So, the next step was to install Office 2000 on the WinGate server. Presumably, I'll be able to publish

* * * * *

And yet another thing put me in a bad mood yesterday afternoon. We have a service contract from Logan on our new furnace/air conditioning system. Barbara scheduled a visit for them to do the spring/summer checkup. The guy showed up yesterday afternoon. I let him into the basement and went back to work. Fifteen minutes or so later, the front door bell rang again. It was him, presenting a work order for my signature. 

Our furnace has a humidifier drum on it. The thing is about 6" in diameter and a foot long, and is covered by a foam filter. The bottom rests in a pan of water, and the drum rotates through the water, wets the foam filter and passes air over it. He had replaced the foam filter, for which they charged us $27. I think that's pretty outrageous. I mean, we're talking about a ten-cent piece of foam rubber. Apparently, our service contract covers labor, but not "consumables" such as filters. Okay, fair enough. But by charging $27 for a small foam filter, they're in fact covering the entire cost of the service call, even though we've already paid for that up front.

* * * * *

Hmm. I see that Amazon and other on-line booksellers are offering PC Hardware in a Nutshell for sale already, although I'm still writing it. The cover is pretty, though. Here's a small image of the cover, and here's a larger image. The only problem is, it lists me as sole author. I mailed my editor yesterday to ask him to make sure that Barbara's name also appears on the cover as author.

* * * * *

I spent some time on the phone yesterday morning with my contact at Matrox. She's sending me one of their new Millennium G400 video adapters. This is a new product for which demand is very high, so eval units are hard to come by. It should arrive in the next week or two, and I'll probably use it in the new personal workstation I'm building.
I've been a fan of Matrox video cards for many years. They're fast, reliable, well-built, rock-solid, and display clean images. I've never experienced any incompatibilities with Matrox video cards, and that's not something I can say for most other brands of video card I've used. Matrox drivers are robust and largely bug-free, again something that's not true of many competing products.

I've gotten behind the curve on video cards--the most recent Matrox cards I have running in any of my current systems are several Millennium II/PCI cards--and so it was time to check out the latest and best that Matrox has to offer. I suspect I won't be disappointed.

* * * * *

I've installed FP2K on the resource server, which has a direct Internet connection. I started the publishing process, and it seemed to be working. I killed it, because I wasn't yet ready to publish. I'll try with this and see what happens.

Later: I started publishing at 9:00 a.m. It's now 9:40, and publishing finally succeeded. FrontPage insisted on publishing my entire site, 8 MB and 40 minutes worth. I suppose that makes sense, because as far as the FTP Publishing Wizard was concerned, it had never published any of this material. I was hoping that it would simply compare file date/time stamps, but I guess that's too much to hope for. At any rate. I'm changing this page and will publish again. I hope it publishes only the three files I'm changing (index.html, thisweek.html, and 0621RTDN.html). We'll see.

Later Still (about 13:00): The smell seemed to be getting worse, so I decided I'd better shut down the suspect system and unplug it. Unfortunately, that system is also resource server (including the Internet gateway for the whole network), and is also the system I was using to publish with. So....

I relocated the modem and phone line from the resource server to kerby, my main workstation. It didn't have DUN installed, so I installed the Routing & RAS upgrade and set up my dial-up entry for BellSouth. WinGate (the Internet gateway software) requires at least SP4, and kerby had only SP3 installed. So I installed SP5 (a beta, no less--when will I ever learn?). While I was at it, I changed kerby's IP address from 192.168.111.161 to 192.168.111.203, which was the IP address of the old machine. All the other machines point to the WinGate server by IP address, so that solved that problem.

After spending a few minutes configuring WinGate in its new home, everything works fine. The birds are singing again. Well, I spoke too soon. All machines on the network can receive mail, view web pages, etc., but none can send mail. I'm pretty sure the older versions of WinGate had an SMTP Service, but this one doesn't. I got it fixed in a few minutes by installing the TCP Mapping Service and remapping requests to port 25. That'd be enough to boggle most people that don't write TCP/IP books for a living, though. According to the WinGate help files, it does this stuff automatically if you choose Custom Install, but not if you choose Express Install, which it recommends. That's pretty odd.

Now to find out if I can publish my web from this machine.....

Later Still (about 16:15): Well, I thought I'd published. Of course, when using the Web Publishing Wizard, it's not immediately obvious where you're publishing to. As it turned out, I'd published, but to a subdirectory rather than to the main directory, so of course the web site wasn't really updated. \

I finally decided to give up entirely on publishing with Microsoft FrontPage and to use a standard ftp client instead. I spent some time trying to get my favorite ftp client, CuteFTP, to work, but I couldn't get it to connect to my site. So I ended up downloading the FTP Voyager program (from the WinGate folks). That worked well. It has a synchronization feature that automatically parses the directory tree on the remote server and compares what it finds there to what's on the local copy. I've synchronized both directions, so I hope my local and remote sites are now identical. Now, to try publishing again...

Later Still (about 17:00): FTP Voyager didn't work out too well. I had the local and remote copies of my web fully synchronized. I then closed the FTP application, opened FP2K, made changes to three files (thisweek.html, 0621RTDN.html, and index.html), saved those changes and exited FP2K. I then fired up FTP Voyager again, opened my ftp site and did a compare directories between the local and remote copies. Scads of files showed up as being out of sync. Oh, well.

Back to using FP2K to publish. It finally hit me that my IP address is a unique identifier for the site. All I should have to do is publish to "216.92.40.142/public_html". And, as it turns out, that works just fine. I published to that and everything appears to be working as it should. FP2K took quite a while to publish, because it thought that many files were different between the local copy and the remote web site. Okay, I've now published, and everything should be in synch. I again updated the standard three files, and will attempt to publish once again. I hope that only those three files get transferred...

 


 

 

 

Search this Site

TTG Home

Robert Home

Daynotes Home

Special Reports

Current Topics


Friday, June 25, 1999

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]


The FrontPage publish operation went exactly as it should have. Only the three changed files were uploaded, and the whole process took only a couple of minutes. Faster, in fact, than publishing with FP Extensions installed ever was. At this point, I have things set up Good Enough. I can edit my pages. Barbara can edit her pages. I can publish. FP keeps track of the changes and the links. All is well.

* * * * *

I talked with Robert Denn, my O'Reilly editor yesterday afternoon. I told him about the burning smell, and the fact that I still can't track it down to a particular component. I told him that what really worried me was that all this stuff is physically wired together on a network. Most things fail-safe. That is, they may blow, but they don't put 110 volts on a circuit designed for 5 volts. But it could happen. 

Robert observed that "that would be a catastrophe". I said, "a disaster perhaps, but not a catastrophe" and asked if he was familiar with Disraeli's observation on those words. Asked to define the difference between a disaster and a catastrophe, Disraeli replied, "If Gladstone fell into the Thames, that would be a disaster. Were someone to fish him out, that would be a catastrophe." 

Here are a couple of other Disraeli witticisms:

  • At a social gathering, Gladstone said to Disraeli, "I predict, sir, that you will die by hanging or from some loathsome disease." Disraeli replied, "That, sir, would depend upon whether I embraced your principles or your mistress."
  • During parliamentary debate on the Abyssinian crisis, Gladstone, looking at Disraeli as he spoke, characterized that country as being "fit only for lunatics and Jews." Disraeli responded, "then Abyssinia seems fit for either of us."

Queen Victoria loved Gladstone and hated Disraeli, probably as much for his political principles as because he was Jewish. But there's little doubt that Disraeli was the better man with the bon mot.

* * * * *

This from Shawn Wallbridge [s_wallbridge@yahoo.com]: 

Hello, I sent you an email a few weeks ago asking about Cable Modems and WinGate. Well everything worked fine and I bought WinGate Standard.

Notice I said worked. Well it's a long story, but I decided to move to NT5 from 98. Everything worked great under 98. Now that I am on NT I am having some problems. Most of them I have worked through. The one problem left is close to the problem you had today with Kerby.

I can receive mail (I assume send as well, but my ISP does not allow mail to be sent from their mail server unless it comes from their dial in) from my ISP account (shawnw@pangea.ca), but I cannot connect to send or receive on my Shaw account. The problem is that Shaw doesn't have a fully qualified name for their mail server (at least not one they will tell me) so I have to tell Outlook to connect to 'mail'. I am guessing that this is the problem. I can do everything else (AFAIK).

Any ideas about what I have to change in NT5P or WinGate to get around this problem?

That's pretty strange. I've never heard of a service provider refusing to provide the name of the mail server to the customer. You should be able to find out the name/address of your Shaw mail server simply by pinging "mail" from the machine that's connected to Shaw.

But the problem may lie elsewhere. You have to configure the clients to point to the WinGate server rather than directly to the mail server. For example, my WinGate server has the private IP address 192.168.111.203. My SMTP Server is mail.lig.bellsouth.net. I POP my mail from mail.ttgnet.com, for which the account name is ttgnet. To set up Outlook on a client to POP and send mail via SMTP through the WinGate server, I must configure Outlook as follows:

Server Information:

Incoming mail (POP3): 192.168.111.203
Outgoing mail (SMTP): 192.168.111.203

Incoming Mail Server:

Account name: ttgnet#mail.ttgnet.com
Password: <my password for mail.ttgnet.com>

The Outgoing Mail Server section is blank on my configuration, because BellSouth does not require authentication to access their SMTP Server. Instead, they restrict access to only people dialed in to their local network. Some (very few) service providers also secure their SMTP servers, either with SMTP-after-POP or directly with a secure SMTP server. I doubt that Shaw does that.

If this setup works for receiving mail, but you still can't send mail, you probably need to set up the TCP Mapping Service on the WinGate server. I have TCP Mapping set up to accept connections on Port 25 (the SMTP port) and a mapping to "Link to mail.lig.bellsouth.net:25 no dependencies"

Hope this helps.

 


 

 

 

Search this Site

TTG Home

Robert Home

Daynotes Home

Special Reports

Current Topics


Saturday, June 26, 1999

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]


I spent most of yesterday writing and on the phone or exchanging email with vendors. When it came time to start my weekly network backup, I realized that my tape drive was in the resource server, which I'd shut down and disconnected. I have other tape drives around here, but none is installed in anything. So, for the first time ever, I did a backup to an optical disk, using Adaptec DirectCD to copy my entire data directory to a CD-RW disk. It's certainly not a full network backup, but it'll do for now.

After dinner, Barbara took Duncan to flyball practice. I was just sitting down to get some more writing done when Pournelle called. We ended up talking for an hour about FrontPage, Allegheny (Agony) Airlines, Buffy the Vampire Slayer, PCExpo, and a bunch of other stuff, including how to tear Princess down to bare metal and rebuild it from scratch. So it goes.

Last night, I finished Tom Clancy's latest, Every Man a Tiger, which he co-authored with General Chuck Horner, and Lawrence Block's latest Bernie Rhodenbarr mystery, The Burglar in the Rye. I have to say that I enjoyed the Block more than the Clancy. I'm sure that Clancy's non-fiction is very popular at the Pentagon and the War College, but the details of theater-level strategy and logistics of Operation Desert Storm simply didn't hold my interest. I wish Clancy would get back to writing fiction.

* * * * *

This from Shawn Wallbridge [swallbridge@home.com]

Thank You very much. Everything seems to work OK now.

Shawn Wallbridge

You're welcome. Glad you got it going.

* * * * *

This from Chris Carson [root@mail.bcsupernet.com; on behalf of; cc [cc@bcsupernet.com]]: 

I've been looking at your disscusion of the Wingate deal.

I think it's safe to say this is a copy of the concept of the Linux masquerade server. I say concept because they obviously work differently.

A masqurade server is probably somewhat harder to set up (I don't find it too difficult) there are a couple of arcane lines to create in several places but that's it.

Once a masqurade server is set up all it does is strip the local IP from the packet then adds the IP it got from the DHCP at your ISP's, when it gets the packet back it does the reverse. These means everything is exactly the same as not going through it no setting mail addys nothing.

The reason this is possible is because of the IP aliasing facility in the kernel. Wingate is some kind of workaround no doubt as I'm pretty sure no MS kernel does that.

The result is that a masqurade server is a little harder to set up but configuration is not required. OK you'll have to tell the inhabitants of your LAN the gateway addy but thats it.

A thought I had. "User friendly _always_ means it has to be broken in some attractive way".

CC

--

Upgrade to Linux...the penguins are hungry!
Chris Carson aka "GreyDeth"
250-248-0142
http://carnagepro.com

Well, the converse, perhaps, given that WinGate has been around a lot longer than Linux. Actually, the difference between the two is in the distinction between two pretty well-known methodologies, each of which has advantages and drawbacks. WinGate is a proxy server, whereas the Linux IP masquerade facility is essentially a Network Address Translator (NAT). There are several third-party NAT utilities for Windows 95/98 and Windows NT. One of them, I can't recall its name, was recommended here by a reader some months back.

* * * * *

This from Chris Carson [root@mail.bcsupernet.com; on behalf of; cc [cc@bcsupernet.com]]: 

Ah it's a proxy server, thanks, that makes more sense. I wonder how NT does IP spoofing though. I don't belive it is a kernel service ... not one I ever found, though it's been a while since I had a windows system, be about 18 months now.

Thanks, I'll be able to make more sense when people ask me the difference.

CC

--

Upgrade to Linux...the penguins are hungry!
Chris Carson aka "GreyDeth"
250-248-0142
http://carnagepro.com

No, as far as I know, NT does not support IP masquerading natively (although if you install the Option Pack you get an SMTP module that runs in an IP masquerading environment. You don't get the smart mailer, though. For that, you still have to run a real SMTP package on Linux or a commercial NT product). I'm not sure which technology Windows 98 SE uses. From the brief descriptions I've read, it sounds more like a proxy server similar to a crippled WinGate clone, although it may be a NAT.

 


 

 

 

Search this Site

TTG Home

Robert Home

Daynotes Home

Special Reports

Current Topics


Sunday, June 27, 1999

[Last Week] [Monday] [Tuesday] [Wednesday] [Thursday] [Friday] [Saturday] [Sunday] [Next Week]


Every Monday, I download the raw usage stats from my server at pair and then run Analog to analyze the log files. I usually don't pay a lot of attention to the logs, but in the process of migrating things over to kerby, I decided to go ahead and download and process the stats today, just to make sure everything worked. While I was glancing over the results, I noticed the Domain Report section, which I don't usually look at. I may start looking at it because there's some interesting stuff here. 

Incidentally, the "#reqs:" column refers to "requests" as opposed to page reads. One page may comprise several "requests"--the HTML page itself, the background graphic, and so on. My site averages about 5,207 page reads per 10,000 total requests, so the actual number of page reads is just over half the number in the #reqs: column. At any rate, here's who's visiting this site:

#reqs: %bytes: domain 
-----: ------: ------ 
9434:  38.33%: .com (Commercial) 
6735:  31.20%: .net (Network) 
2641:  10.77%: [unresolved numerical addresses] 
1078:   2.74%: [domain not given]
 698:   2.71%: .ca (Canada)
 624:   2.70%: .edu (USA Educational)
 319:   1.37%: .nl (Netherlands)
 268:   1.12%: .au (Australia)
 358:   0.95%: .ie (Ireland)
 236:   0.87%: .org (Non-Profit Making Organisations)
 215:   0.86%: .mil (USA Military)
 206:   0.75%: .us (United States)
 233:   0.72%: .arpa (Old style Arpanet)
 143:   0.63%: .uk (United Kingdom)
  82:   0.53%: .de (Germany)
 101:   0.47%: .za (South Africa)
  85:   0.35%: .hk (Hong Kong)
 129:   0.35%: .se (Sweden)
  68:   0.28%: .gov (USA Government)
  49:   0.27%: .jp (Japan)
  56:   0.22%: .pg (Papua New Guinea)
  60:   0.20%: .fr (France)
  42:   0.19%: .be (Belgium)
  29:   0.14%: .nz (New Zealand)
  32:   0.11%: .it (Italy)
  17:   0.10%: .ru (Russia)
  44:   0.10%: .ch (Switzerland)
  45:   0.09%: .co (Colombia)
   2:   0.06%: .pl (Poland)
   3:   0.06%: .lt (Lithuania)
   5:   0.06%: .bg (Bulgaria)
   9:   0.06%: .mx (Mexico)
  21:   0.06%: .br (Brazil)
  14:   0.05%: .ua (Ukraine)
  40:   0.05%: .my (Malaysia)
   3:   0.05%: .pt (Portugal)
   6:   0.05%: .es (Spain)
   3:   0.04%: .gr (Greece)
   3:   0.03%: .tr (Turkey)
   4:   0.03%: .il (Israel)
   5:   0.03%: .fi (Finland)
  10:   0.03%: .dk (Denmark)
   6:   0.03%: .sg (Singapore)
   3:   0.03%: .np (Nepal)
  11:   0.03%: .cr (Costa Rica)
   5:   0.02%: .in (India)
   2:   0.02%: .md (Moldavia)
  10:   0.02%: .at (Austria)
  12:   0.02%: .is (Iceland)
   8:   0.01%: [unknown domains]
   9:   0.01%: .th (Thailand)
  11:   0.01%: .ar (Argentina)
   7:   0.01%: .ph (Philippines)
   3:   0.01%: .uy (Uruguay)
   5:   0.01%: .sa (Saudi Arabia)
   4:        : .id (Indonesia)
   3:        : .lk (Sri Lanka)
   8:        : .kr (South Korea)
   1:        : .no (Norway)
   1:        : .ee (Estonia)
   1:        : .hu (Hungary)

* * * * *

This from Jim Griebel [jgri@earthlink.net]: 

The Internet Connection Sharing feature on W98SE doesn't seem to be a proxy; at any rate the Netscape client I run on the Linux box doesn't need to be set up for a proxy. (Yes, to the horror of Linux users everywhere, I'm accessing the Internet from the Linux box via the Windows box.) I gave the Linux box the Windows machine's IP as the default gateway and told it which DNS the ISP uses, and that was pretty much that. (Other than getting Linux to quit demanding a connection to the DNS at odd intervals.) ICS provides at least some DHCP services for clients, but I haven't really put that to the test yet. 

The main drawback of ICS that I can see is that it insists on a fixed subnet -- 192.168.0 -- which would mean it won't work with some setups, unless maybe you put a second network card in the gateway machine. 

While going through all this networking stuff (I upgraded the home network, all twelve feet of it, to 100MB Ethernet last week) I noticed something odd. I had a problem with the network card in the W98SE machine -- it would send, but not receive, which turned out to be related to buggy IRQ steering, or something like that -- and while it was vainly trying to find the rest of the network it assigned itself an arbitrary IP address (169.254.something) and started looking for other machines on that subnet. This didn't bother me, but I wonder what would happen if you just happened to have that subnet and a flaky Ethernet card somewhere. 

Thanks. I'm currently running Win98 (aka Win95 v 1.1) rather than Win98 SE (aka Win95 v 1.11), so I haven't had a chance to look at Internet Connection Sharing yet. From what you say, it sounds like a simple software NAT. As far as the 192.168.0 subnet, that's probably a reasonable thing to default to, although I don't much like it being a forced setting. As far as Windows assigning itself an arbitrary IP address, that's pretty poor practice to say the least. 

* * * * *

This from matt horrocks [matts@burgoyne.com]: 

That page about how you made explosives is cool. How you made black powder at 11 is amazing and then nitro at 12 hehehe!!! Were did you learn to make things like that. Its hard to find the chemicals now. good story, bye.

Well, finding out how to do it is the easy part. Living through your first attempts is what's tough. As I recall, I visited our local library and started reading Army Field Manuals that explained such things as unconventional warfare and field-expedient munitions. Everything you really needed to know was in there, and it was authoritative. I also picked up a copy of The Chemistry of Powders and Explosives.

I doubt that the chemicals are all that much harder to come by now than they were then. Since Oklahoma City, of course, everyone knows about ANFO (ammonium nitrate/fuel oil). Even so, the last time I noticed you could still buy a 50 pound bag of ammonium nitrate at the garden supply store, and all the fuel oil you wanted at the diesel pump. That's the problem. All the chemicals needed are common ones, and all have many legitimate uses. 

ANFO isn't much of an explosive, really. Technically, it's a high explosive because it detonates (or can be made to detonate under proper conditions), as opposed to, say, black powder, which does not detonate but simply burns very, very quickly. But ANFO has such a low velocity of detonation that it has very little brisance, or shattering power. Instead of shattering, it heaves. About the only military use for it is in things like road cratering, where you want heaving action more than shattering action.

For someone who wants to manufacture real military-grade explosives, such as RDX, it's not all that much harder to obtain raw materials. Nearly everything you need for almost any common explosive is readily available as an industrial chemical. I can go out today and buy all the ammonia, formaldehyde, sulfuric acid, and nitric acid I need to make a pound (or a ton) of RDX. It's undoubtedly harder for a kid to gain access to the raw materials, but that's no bad thing.

As I've said repeatedly in the past, I strongly urge anyone who's considering playing with explosives to think again. It's bad enough that you risk your life by experimenting with them. But even if you're careful and survive that, the courts have no sense of humor about explosives anymore. When I was blowing things up 30 years ago, people just shook their heads. No one thought it was a matter for the police or the courts. Nowadays, you're likely to get caught, go to jail, and pay a huge fine. That's if you don't kill yourself first.

* * * * *

This from matt horrocks [matts@burgoyne.com]: 

I have been wonder about explosives and chemical reactions. That stuff is neat I think. Im 14 years old and I am going into high school. I want to learn some chemical stuff there. When I was 10 I saw a few guys setting up a firework show. At the time I did not know of chemicals or anything like that. So I looked around and asked questions. I learned a little bit about it and I thought I wanted to be a pyrotech guy. Well I fixed up some fireworks to explode and stuff like that or got dry ice and put it in plastic bottles. Its was fun. But I still didnt know much about chemicals. So I got on the internet and look around, I found neat stuff. Now I think I will learn some more stuff about everything im intrested in.

Okay. Once again, I'll warn you. Don't do it. Join the Rocket Club or find other ways to enjoy pyrotechnics. But don't mess around with making them yourself. If you do, you'll almost certainly hurt or kill yourself or other people. This stuff is not something for kids to mess around with. Come to that, it's not something for adults to mess around with. Unless you know exactly what you're doing--and you don't--compounding your own pyrotechnics is a quick way to the hospital or the graveyard. Even people who know what they're doing get killed all the time. Believe me on this. 

I once came within literally five seconds of being killed. I was working on a device and had forgotten something I needed. It was behind the concrete block wall that I intended to shelter behind when I set the device off. I walked behind the wall to get what I'd forgotten. The device I'd been working on a few seconds before spontaneously detonated. If I hadn't been behind that wall, they'd have had trouble identifying what was left of me. And I knew what I was doing. As it turned out, the problem had been that I used some chemicals that were supposed to be of a certain purity and turned out not to be. The impurities were sufficient to catalyze the reaction. It wasn't my fault at all. But that wouldn't have helped me if I hadn't been behind that wall. Trust me on this. DO NOT ATTEMPT TO MAKE YOUR OWN EXPLOSIVES. 

* * * * *

This from Guntis Glinavs [gglinavs@serix.com]:

Just want to compliment you on your site and efforts - interesting reading on lots of computer problems I hope I never have to solve (I have enough different ones of my own)

Re: W98 SE ICS

I'm pretty sure the ICS module is a NAT - probably based on a program created by NEVOD Inc. These people had NAT products for W9x and NT that seemed to be the only ones to allow Quake II and Blizzard's Battle.Net related games to work from a private network -> NAT -> cable modem. I've had a 5 user license from them for over a year and it has been bulletproof - transparent to the clients, provides automatic DHCP, seems fairly secure (so far) and it works - what more could I ask?

In February they had a notice on their website that Microsoft had bought them out and that the web site's help/support files would only remain in place till April. They emailed out the latest version of their program and then disappeared!

They subsequently advertised for beta testers for a W98 module so it would be safe to assume that NEVOD's NAT1000 has been incorporated into our favorite OS. If it is as capable as NAT1000 most people should be pretty happy.

>From my experience with WINGATE, the NAT solution seemed a whole lot easier - no port configuration, no security hitches ... Apparently WINGATE (at least in earlier incarnations) has/had some severe flaws when setup on a cable network using its defaults. These included providing conflicting DHCP services OUT to the external network and allowing relatively easy email/irc spoofing from outside locations. I know that Rogers_@Home was unhappy with WINGATE users and various IRC servers tended to ban entire IP ranges of cable users due to the problems caused by misconfigured WINGATE servers. 

On a similar note I wonder if you've seen some of the new thin server (?) software products that are appearing. One in particular seems interesting - Network Concierge (http://www.nc4u.com). This looks like a relatively easy and hassle free way of setting up a LINUX network server to provide file and Internet features on a low cost hardware system without having to learn all of the arcane Linux spells and incantations. Any comments?

Keep up the good work.

G^2

Guntis Glinavs
Nanda Devi Computing
London, Ontario
gglinavs@serix.com

Thanks for the kind words. From what I've heard from you and other readers, it seems reasonably certain that ICS is indeed a NAT. I've not experienced any of the problems with WinGate you mention, but then I don't have a cable modem (or ADSL, more's the pity). I'm not familiar with the thin server stuff you mentioned, but I'll take a look when I get a spare moment. There's no doubt that installing Samba for file sharing is well beyond the abilities of many would-be Linux users, and setting up an Internet server (particularly mail) is beyond most. I can see that a product like you mention might be very attractive to quite a few people.

 

 

Copyright 1998, 1999, 2000, 2001, 2002, 2003, 2004 by Robert Bruce Thompson. All Rights Reserved.