For example, if I search for 'Eclipse' while wearing the 'Software Developer' hat, I should get Eclipse IDE related links before links related to the astrophysical phenenomon. If even I was interested in the later, results I get back should be different depending on whether I am wearing a Physicist's hat or a Photographer's hat.
That example is rather silly, though. If you're searching for the Elipse IDE, obviously you can just search for eclipse ide or eclipse java development, and if you're searching for astronomical eclipses, you'd search for something like lunar eclipse.
What would be useful is an "I'm not shopping for anything" option. There are many sites that get in the way in Google's search results when you're trying to research certain consumer electronics, for example.
JavaHMO is an application that allows you to use the Tivo Home Media Option on platforms other than Windows and OS X. (It works on those platforms, too.) The RedHat packages available on the SourceForge site install fine under Debian after being converted by alien, but the RedHat init script doesn't work, so I wrote an init script for Debian.
The init script runs JavaHMO as the
javahmo user, so you must create the user before using it.
# adduser --system --home /usr/share/javaHMO --shell /bin/false --group javahmo
Update: I created a proper Debian package for JavaHMO that doesn't require using alien to convert the RPM.
I've been using sender address verification callbacks for a long time. It helps eliminate a lot of spam by checking if the sender's address is deliverable. Unfortunately, there are a number of systems that send mail with an invalid envelope sender. These are often generated by scripts on a web server where the sender defaults to email@example.com. There are also a number of misconfigured mail server, mostly IMail installations, that do not accept messages with null senders. This not only prevents their users from receiving bounce messages, but also prevents sender address verification from working.
Up until yesterday, I've rejected messages at RCPT time that fail sender address verification. Trying to deal with the number of false positives for a significant number of users has proven to be too dificult. So I decided to continue using sender address verification, but incorporate the result into an overall SpamAssassin score.
Andrew, on the exim-users list provided a helpful Exim ACL snippet which I modified a bit and came up with the following:
acl_callout_test: warn set acl_m6 = TEMP accept verify = sender/callout=60s,random set acl_m6 = OK warn set acl_m6 = FAIL acl_check_rcpt: warn acl = acl_callout_test warn message = X-Sender-Verification: $acl_m6
This adds an X-Sender-Verification header which I then check for in SpamAssassin.
header POSTICA_SENDER_ADDRESS_FAIL X-Sender-Verification =~ /FAIL/ describe POSTICA_SENDER_ADDRESS_FAIL Sender Address Verification Failure score POSTICA_SENDER_ADDRESS_FAIL 2.0 header POSTICA_SENDER_ADDRESS_TEMPFAIL X-Sender-Verification =~ /TEMP/ describe POSTICA_SENDER_ADDRESS_TEMPFAIL Sender Address Verification Temp Failure score POSTICA_SENDER_ADDRESS_TEMPFAIL 1.0
I may have to tweak the scores, but so far, it's working pretty well.
Back in 1999, after the Lincoln Navigator was introduced, I joked, "Next year, people will be driving semis around town." Looks like I was only off by a few years.
MSNBC has a picture and
details on the 14,500-pound behemoth.
But does anybody really need a vehicle that is nine feet tall, eight feet wide, 21 1/2 feet long and gets about seven miles on a gallon of diesel?
"Anybody can lease a Hummer now for $500 a month. Soccer moms are driving them," he said. "This is not a soccer mom's vehicle."
We'll see about that.
Mailgraph is a
great tool for monitoring your mail servers. It generates graphs showing how
many messages are passing through your servers as well as the number of spam
and viruses blocked. Here's a sample graph:
Apparently, it is now possible to inhale alcohol. I'm not sure why anyone would want to, though. I can't imagine going into a pub and having the bartender put down a pint of your favorite vaporized beer.
That sounds about as likely as someone listening to a crappy version of their favorite song on their mobile phone.
According to Derek, hrefs such as
<a href="https:/path"> should be valid and link to the same site using
https as the protocol.
This doesn't work in Mozilla, which converts such links to
https://path/, nor in links which interprets it as
Interestingly, it does work in Mozilla if the link is the same protocol as you're currently using.
If it worked, it would make web development easier. You wouldn't have to use tokens in html templates for the server name to avoid hard coding in the production server's domain. I'm sure there are other's who've used output buffering hacks to rewrite relative links.
So, what's the deal? Are all web browsers broken?
When using an external server or servers to filter your mail, you want to make sure that spammers and worms cannot bypass the filtering servers and deliver unsanitized messages directly to the destination mail server. In order to do this, the destination MTA must check the IP address of the server trying to deliver a message and verify that it is one of the allowed hosts.
If you only have one domain on your sendmail server, or all domains use the same filtering hosts, this is fairly easy to accomplish using either a firewall, TCP wrappers, or a number of not-so-elegant entries in your access table. If you host a number of domains, these methods may not be acceptable.
I have written some sendmail rules to restrict delivery on a per-domain basis. Groups of scanning hosts, known as a scangroup, can be setup and each domain can belong to one scangroup. If a domain does not belong to a scangroup, messages to that domain are accepted as usual.
Two new maps need to be created,
scandomain map lists each domain that uses a scangroup and
the IP addresses of each host in a scan group. The left hand side is the
domain and IP address, respectively, and the right hand side is the scangroup
name for both maps.
The rules which reject unauthorized delivery are added to the
ruleset so the rejection occurs after each
RCPT TO: command. Because the
mail server may also be used for message submission by MUAs, it accepts
messages from authenticated users and IPs explicitly allowed to relay in either
relay-domains file or
To enable scangroups, add the contents of
scangroups.mc to the bottom of your
sendmail.mc and regenerate your sendmail.cf. Don't forget to run makemap after
setting up your
If you're looking for a service to filter spam and viruses from your mail before they get to your mail server, please check out Postica, which I developed.
Updated - 5/2/2005: I updated the ruleset to reject messages with a temporary failure so that mail doesn't accidentally get rejected if DNS is being changed, and still points directly at the destination server.
Over on TJ's Weblog, they picked a company planning on opening a spa in Slovakia as one of their top 3 business plans.
TOP 3 - comes in MAGMA ZAFÍR s.r.o. that aims to develop a hot water spa in Slovakia. Central Europe has been less then well served with wellness facilities so far and has the advantage of cheap labor costs. A world-class spa for clients from Germany and Austria could easily fill a market void.
I'd be interested in reading their business plan. Slovakia already has a number of spas, and Piestany has attracted foreigners from around the world to its spas for a long time, even during Communism.
Slovaks, in general, see spas as places for the elderly and infirm, but that may start to change. When I was in Slovakia a few weeks ago, there was an article in a magazine describing a visit to a spa by a young woman. She hadn't been too thrilled about her assignment to visit and write about her experiences at a spa. She figured she had twenty or thirty years before she should start visiting spas. She was delightfully surprised by her experience and concluded that visiting spas was a great way for young professionals to relax.
I think there could be a good opportunity in marketing spas to Slovaks between 20 and 40.
There was one ad in a recent edition of Time Magazine that caught my eye. We've been trying to figure out how to best market Postica since officially going into production in July. I'm pretty skeptical about the effectiveness of advertising in general. Since we're bombarded with so much of it, I think most people simply ignore it. But every once in a while, some advertisement will catch my eye. I don't know if the fact that I read it means that it actually increases sales of the product advertised, but it obviously can't be worse than an ad that nobody reads.
The full-page ad in Time was for men's white dress shirts. Just plain, white button-down shirts that I buy every 5 years from Ross when the one I bought five years previously no longer fits. I had never heard of the brand; it was a man's name (two first names, I think). It was one particular shirt, rather than an ad for the brand in general.
What made me read the ad was simply that I wouldn't expect to see such an ad in Time Magazine. I wonder how sales of that shirt are doing. I believe one could order the shirt in question online or by calling an 800 number, similar to ads for overpriced cds composed of the Best Love Songs of Some Period within Some Genre or "collectible" coins that aren't legal tender.
Does placing an ad where potential customers wouldn't expect to see it help sales?
Here's how to print some documentation as a book. The goal was to print two pages per sheet double sided without having the luxury of a printer that supports duplex printing.
So we use
psselect to pick the pages we want to print.
Then, we use
psbook to rearrange the pages so they end up in the right order
in the finished booklet. The
psnup utility is used to to print two pages per
sheet, and finally,
pstops selects just the odd pages in reverse order.
In the second step, you feed the pages back through your printer using the
manual feeder. On my printer, a LaserJet 1200, the manual feeder pulls page
from the top, the same as the tray. That is why I used pstops to rearrange the pages.
If your manual feed pulls from the bottom, you can print the odd pages in the normal order.
psselect -p34-85 op.ps | psbook | psnup -2 | psselect -o | lpr
Now, we just put the pages in the manual feeder and print the even pages.
psselect -p34-85 op.ps | psbook | psnup -2 | psselect -e | lpr
P.S (Ha!) I think
pstops can do everything that
do, but I didn't bother learning the syntax.
The state is that great fiction by which everyone tries to live at the expense of everyone else. - Frederic Bastiat