IEWatch Redux

Several folks chimed in on my post about IEWatch last week. I also submitted IEWatch a sales support request asking for a Firefox version, and they said that there are no current plans, but “We might consider writing a compatible version if we receive enough customer requests.”

One of the compelling aspects about IEWatch is that it puts the connection information right into your browser window. At the cost of serious screen real estate, but we all have large monitors now, don’t we? None of the alternatives presented in the comments do this: even the Live HTTP Headers plugin for Firefox opens a separate window. In that window it dumps all the request and response headers in a fairly disorganized fashion, giving the user both too much information and too little. IEWatch neatly organizes the request and response headers, content, cookies etc. in the tabs across the bottom of the browser window.

Compelling Internet Explorer Add-in

If I were running Windows, this would actually make me use Internet Explorer more. The IEWatch gadget gives you an ?underwater screen? that shows every HTTP transaction made when loading your pages. Great for troubleshooting and debugging, and you don’t have to run tcpdump or sift through mounds of traffic to piece together your web transaction flow. It almost makes up for IE’s atrocious error reporting, and because it shows every 404 or Redirect response that occurred has helped me tremendously in troubleshooting some weird edge case issues at customer installations.

Screenshot of IEWatch

So, dear Lazyweb, does anything like this exist for Firefox? Having this for Safari would be even better, but an IEWatch-like add-in would have the power to make me switch browsers on my Powerbook.

New SSL Certificates, now with Green which is More Safer!

As noted in The Register, Verisign teams up with Microsoft to enhance the user experience of Internet Explorer 7 when browsing SSL-protected sites. Verisign will sell High Assurance certificates to sites that pass a more stringent identity verification than is currently the norm. When it encounters such a certificate, IE 7 will turn the address bar green in addition to displaying the usual padlock. A Phishing Filter (Philter?) turns the address bar red when the user accesses a known phishing site.

While I hope that they include enough visual cues for the red/green colorblind among us, I don’t dislike this idea. Is it a scam? Not necessarily. Details about what a High Assurance or Extended Validation certificate actually comprises are scarce, but it’ll probably take the form of a certificate attribute that Verisign will set on these mo’ expensive, mo’ better certificates. Such an attribute can be set by any CA, parsed by any browser and can be ignored by the enormous installed base of credit card wielding, revenue generating users of older browsers. Whether or not a company drinks the Microsoft/Verisign Kool-aid, they hopefully won’t stand for breaking backwards compatibility. On the other side, it’s the responsibility of the Certificate Authorities to only set this attribute on their mo’ better certificates, for which they in turn can charge mo’ money.

This whole thing ties into a new concept of Trust. The situation is not black and white anymore. Trust is the new green. Or yellow, or red. You can get a cheap certificate by proving that you can ping an e-mail back and forth to the CA. This shows them that you have access to e-mail on the domain, which is good enough for them as an (automated) identity verification. Whether said domain is practically undistinguishable from that of an actual business falls outside this check. One would dearly hope that applicants for a High Assurance certificate undergo more scrutiny than that.

Earlier this month at ApacheCon, I attended a very interesting talk by Lisa Dusseault about Federated Identities. As she talked about rate-limiting the creation of centrally verified identities to thwart spammers, she came up with the Fifty Dollar identity. The knowledge that the party you are talking to has a non-trivial sum of money behind their identity record might positively affect the trust you place in that identity. I see much the same happen with this new server-side certificate paradigm: cheap normal certificates you trust a little, and mo’ Green mo’ better certificates you might trust more. So far, browsers have not given us any idea about the quality of a site’s certificate. It’s either trusted, or the browser puts up a slew of scary dialogs. The red/green address bar might bring some nuance to this concept and put a more human face on the concept of the identity of a web site.

An iPod is a Harddisk

David Lazarus writes in the San Francisco Chronicle how an identity theft suspect was apprehended with his victims’ personal information stored on his iPod. David, an iPod is a hard disk. You can store any type of data on it you want, including music and video. This is not different from a floppy, CD or USB Keychain drive, just larger. And I suppose it has a recognizable brand name that one can hang a column on.

Thank You For the Memory

Early this week, I went ahead and maxed out the RAM in my Powerbook. I now have 2Gb and it just flies. With the default 512Mb it was swapping a lot, especially after a couple of days of opening and closing applications. Especially Safari was hit very hard. Now, switching applications takes no time at all. For under $300, I should have done this months ago. What a difference.

CodeCon 2006

I am currently attending CodeCon 2006 in San Francisco. Interesting talks, as usual. Really atrocious folding chairs that look OK at first glance, but then you sit on them for an hour and realize everything down there is numb.

Three talks on Friday about fighting Malware/Spyware/Phishing. The first two, on Daylight Fraud Prevention and SiteAdvisor had in common that they were about commercial offerings and did not include any code. TRUMAN, The Reusable Unknown Malware Analysis Net, is available under the GPL.

SiteAdvisor combines automated analysis of potentially malicious websites with “a bunch of guys in India??? to produce a rating of the malice these sites wreak on the user’s computer. They put a red/green status indicator in the browser status bar, and inject red and
green icons in Google search result pages. This is pretty interesting and I’d like to see where this goes. The presenter seems to have a good grasp of the fact that they don’t know all the consequences of offering this service on a large scale, how it will be abused etc. Watch that
space.

The other two talks on Friday were about the P2P space: VidTorrent is a P2P streaming application out of the MIT Media Lab. Finally, I think Localhost essentially slaps a file system onto BitTorrent.

The presentations this year seem to have a noticable practical slant. These are actual products that are shipping or close to shipping, not vague unpractical ideas hatched in someone’s basement and destined to never venture beyond.

Everything in Subversion

I finally set up a small Subversion repository on my local server. The box now has a HD without holes in it (for the moment), so perhaps the repository backend will stick around for a while.

Besides my private software projects, this repository will hold the /etc and kernel configs of my local boxes and this blog. That in turn should allow me to start making modifications to the CGI and the unfortunate default templates.

A couple of To-Dos: I’ll need some checkin hook scripture that pushes the entries out to the server as I commit them, and the CGI uses the file system last-modified date to determine publishing date on the pages. That’s real elegant, but not tremendously robust, especially when the content eventually goes through svn, an export, rsync etc.

FreeBSD and Wireless PCI Card

I’m now running my main development PC directly on the wireless
network. This took two trips to CompUSA (three if you count my initial
survey of the products on offer), upgrading to FreeBSD 6.0 and a
custom kernel build, but it is a going concern now. Very exciting, I’m
sure. It is for me, since now this box does not have to live behind an
old Powerbook with a chafing hard disk sharing its Airport connection
in NAT mode, which eliminates one noisy beast from my office.

Initially, I came away with a LinkSys WMP11 card that would cost me,
after two rebates, the grand sum of five dollars. I had googled for
this product and found out it is based on the well-supported Prism
chipset, so I took one home. However, the card I had bought turned out to be a new revision, with a completely different chipset that the
FreeBSD developers had never heard of! Thanks Linksys for springing
that on the unsuspecting public. Of course there are Windows drivers,
so the unsuspecting public still has something to play with. As, it
turned out, did I. I turned to the NDIS wrapper support and found out
that this only came with FreeBSD 5: time to upgrade the box from
4-STABLE. After a fresh new install (the only thing building world did
for me was make every program I started dump core… boy was I glad
that I had backed up my home directory before I started on this
track), I got the NDIS driver to work (great hints by Dannyman),
but unfortunately it did not support WEP. Why do we care about WEP?
Isn’t it completely uncool to rely on WEP these days? Actually, it is
in a way. In the same way that the lock on a bathroom stall door
prompts most people to try the other stall, I assume the
1337 H@x0R is going to hook up to the unencrypted network first and I can see up to six open networks from my second bedroom.

So, I pulled the Linksys card, took it back to CompUSA and got a
Netgear WG311T instead. This card is directly supported by the ath(4) driver… or so I thought. Of course, this was also a newer revision of the chipset, but at least Netgear didn’t switch sources on me. Support for this, the 5212 chip, was only in FreeBSD 6.0, so another upgrade was in order. However, everything works now. Phew.

Why ZeroConf is Good

Just got a new printer, a little laser job that costs about one-twenty-fifth of the first LaserWriters. And it can do color, too. I now have it set up on our little home network, so that both the Macs and the Windows machines can print to it. Mac setup was done in five minutes, including the time it took to find an extra power strip for the JetDirect box. Windows setup took about two days….

As a long time Mac zealot, the five minutes doesn’t surprise me. Two days to get the Windows side running, however, is disappointing in a number of ways. First off, this product is targeted at the SOHO market: it will typically land in situations without a knowledgeable IT person. Secondly, I hate playing uncompensated IT guy. If I am to put up with this ****, I’d like to see some $$$$ out of it. Finally, there still those who will maintain that Windows has now caught up to the Mac and that Apple has no usability advantage anymore. Sorry folks, five minutes versus two days does not look caught up to me. And yes, I am a zealot.

OK, let’s do the full ESR on this. The network has a FreeBSD box with a DHCP and DNS server and a Win2k workstation that shared the old inkjet printer. There are two Windows XP Pro laptops, a Powermac G4 and a Powerbook. The JetDirect box gets its IP address from the BSD box, its configuration web page shows up immediately in Safari on the Powerbook and I give the printer a name (ShangriLaserJet), turn off IPX and AppleTalk, all that good stuff. Actually printing to the printer requires a driver install from CD, which does not present any problems.

Now, Windows. The CD has an installation wizard, which finds the printer on the network in about a minute. It probably does a MAC address search, or has a Rendezvous browser (although that wouldn’t take that long). Click, click, complete wizard and I now have a printer entry in my Windows. Great, except it can’t print. Turns out that it made one of those weird backwards Windows constructions where a TCP/IP connection to the printer is in fact at the same level as a direct parallel or USB port connection. In Windows-speak, a ‘network printer’ is one that’s accessible through one of those \\server\printername constructions. It also turns out that the installer used the printer’s name rather than its IP address for the port. Where did the installer get that name? Must be from the printer itself, not from the DNS because my DNS server doesn’t know about “ShangriLaserJet???. That’s a pity, because now the PC can’t find the printer.

So, I add a fixed address for the printer to the DHCP server on the BSD box, and an A record (and IN record) to the name server configuration. I power-cycle the JetDirect interface, and now the printer can be pinged by name from the Windows laptop and I can print to the printer. Rejoice!

Oh wait, not quite. On the other Windows XP machine, we’re running as a non-privileged user. This in itself causes no end of hassles, as numerous software packages (cough cough Autodesk cough) don’t really deal with running without administrative privileges. On the printing front, however, it turns out that an unprivileged user doesn’t have access to that strange looks-local-but-is-actually-a-network-port. I open up the port properties, Security tab and grant all possible privileges to every account on the box, then add specific blanket privileges to that non-privileged account. All this according to the first principle of Software Security Management: when it doesn’t work, open that bad boy up. Second principle also applies: totally forget to close the hole again when opening it doesn’t work. Which it doesn’t. Solution to this one: install the driver on the final, Windows 2000, PC and share it so the unprivileged user on the XP box can access the printer. We’re up and running.

So, could I have done this without knowing what I know about networking? No way. Must I run my own DHCP and DNS server on my five host home network? Not really: I can just use IP addresses to access the printer from each host. However, the fact that the printer installer uses the Rendezvous hostname of the printer when creating its Windows printer port is severely broken behaviour: it should at least do a reverse lookup on the IP address and figure out that (on most SOHO networks) the printer IP address won’t resolve. Using IP addresses in the local/network printer port would have made this work out of the box, but it would fail as soon as the DHCP address of the JetDirect changed… that’s why Zero Configuration Networking was invented and why this technology needs to come to Windows: otherwise, Fry’s Electronics is going to get a whole lot more printers back because their users couldn’t get them to work on the network.

As for the other problem, I actually hope I’m overlooking some feature in Windows that would allow that unprivileged user access to a local/network printer port. One should not have to run as Administrator in order to access a TCP/IP printer. What if the network doesn’t have a spare PC sitting around that can take the role of print server? This is another thing SOHO users shouldn’t have to put up with. Or maybe the average Windows user knows a heck of a lot more about security and access privileges than I do. Or they just run as Admin.