Sunday, February 26, 2006


A lot of sites recommend blocking spammers with .htaccess. For those who want to get more out of the file than just blocking certain spammers, see this .htaccess tweaking article.

Tuesday, February 21, 2006

PRStorm For Sale Again

After Steve bought PRStorm in October, he told Michael Pollitt, "I hate spam and never intentionally will distribute any spam." Now that he has made tons of money off of distributing spam, he is making good on his "get rid of the PRStorm site very soon" plan. Though not by shutting it down, he is selling it on eBay again.

Thursday, February 02, 2006

Regex Spam Fighting

If you are fighting wiki spam, you likely have found regular expressions are a common method to block spam. They are of course useful for a lot more than just blocking URLs. You can find regexs in a lot of places. But for this post I am sticking to antispam uses.

John Walling uses keyword blocking. I worry about false positives with this method, but if careful with your regexs and you think about the keywords carefully hopefully that won't be a frequent problem. For example, blocking cialis also blocks specialist. But there are words like tramadol or hydrocodone that aren't likely to be necessary unless you are running a medical wiki.

You can also block spam based on patterns commonly found in spam. This is used in blocking CSS Hidden Spam. This doesn't catch all spam, but it will catch all spam using this technique whether or not any keyword or URL would have caught it.

Well, the main purpose of this post was to let you know about's regex tutorial. I saw lots of stuff there I didn't realize was possible. Regular expressions are very powerful tools and you don't need a lot of experience to take advantage of some simple rules. The more people that can come up with their own spam blocking methods the worse off spammers will be. No single method (except locking down your wiki) is going to solve the problem.

Wednesday, February 01, 2006

IE7 Beta 2 Preview Review

Yesterday Microsoft released a public preview beta of . It is nice and simple, but everything has moved and it is a very non-standard UI. I have never used , but the new layout reminds me a lot of the screenshots of NS 8 I have seen.

Every major browser I have tried has the stop and reload buttons next to the forward and back buttons. They make sense there. Things that are going to be used frequently should be in a single place that is easy to find. In this beta of IE 7 the reload button and the stop button shift over to the end of the URL bar. But that isn't the edge of the screen, there you have a dropdown search like and . That is a good place for the search bar, but makes hitting the reload and stop buttons awkward. What is convenient about this placement is the Go button that I always disable finally has a place that doesn't bother me, before a page is loaded it takes the place of the reload button.

The thing that is most strange is the menu bar being missing. It really isn't any loss since everything in there can be accessed by the new buttons, but they are going to take a lot of getting used to. They also take up a lot of the tab bar. There is an option for a classic menu, but by placing it below the URL bar I just can't stand it.

I like that when you hover a tab you not only see the page title, but the address. But I think they need to truncate the address at some point. There is room since it can display from one side of the screen to the other, but addresses with lots of parameters looks bad taking up the entire width of the screen and likely get truncated anyway if they are that long.

I also like the new floating Favorite, Feed, History sidebar. That is a big improvement for screen realestate. Often I see users with that bar open all the time to either History or Favorites even in Firefox. Now that it collapses when you are done it is much nicer. It can be pinned open like before, but this is a great default.

They have added a Phishing Filter similar to the Google Safe Browsing extension for Firefox. I don't really know how well either works, I don't normally stumble into phishing pages so I have yet to see the Google extension warn me of anything except their test page. But in theory this is a great improvement that hopefully will make people less likely to fall for scams. But will these filters make people more likely to fall for scams that sneak through? I would say yes, except that most people that would fall for these attacks now don't know what to look for and would have been fooled anyway. So some protection is better than none.

I found a few rendering bugs without really trying:
Any heavy tab user knows at a certain point in Firefox, tabs just can't get any smaller and are shoved off the edge of the screen. Currently with FF they are no longer reachable through the tab bar. IE will hit that overflow point sooner because some of the tab bar area is taken up with the toolbar buttons. MS seems to have come up with a solution that I haven't seen mentioned at Mozilla. They shrink the edge tabs and use them as a kind of scroll button. Click an edge tab and you switch to that tab and the bar shifts. That is an interesting way of doing it, but not great for power users. And what other kind of users are likely to have that many tabs on purpose? The problem is not simple though, Firefox developers are currently working on figuring out a good way to allow users to access those spill over tabs.

I like the way IE handles the tab close button. They only show it on the active tab and it is grey until you hover your cursor over it so it is not distracting. I never used the close button at the end of the tab bar in Firefox anyway, I always middle click (or double click) and middle clicking also works in IE 7. Firefox developers are experimenting with tab close buttons too, I hope they take a good look at what MS did.

Next is feature already in IE that I just realized, the right click context menu has a big improvement for inexperienced mouse users over Firefox. I have watched older people who just can't get the difference between right and left mouse buttons. By right clicking in Firefox and blindly choosing the first option you open a new window for every link clicked. That makes the back button useless and ends up using a ton of memory. IE's first option opens in the current window/tab. I and many others would probably not like if FF changed the context menu, but the usability improvements that would be for some users would be worth it.

There apparently have been a bunch of CSS fixes which is good. This should really help cross browser site design, but I am sure we will still have to work around many things and of course still support IE 6 for a good while. The best page rendering fix I noticed was that PNG alpha transparency finally works without any hacks. I don't normally use IE so it is hard to tell, but it seemed faster. It certainly was faster than Firefox.

When programs all start coming up with unique interface layouts, users are going to suffer. It may be true that usability tests show the new interface is easier to use, but if every program comes up with their own better way, users will be constantly confused how things are organized. Firefox developers need to keep this in mind too. So far they mostly have made good decisions, but now with what Microsoft has done and their constant strive for a simpler browser I fear they will head in this direction too.

Update: Ted Drake has a post with a bit more detail of the CSS improvements. He also points out that you should realize what you are doing if you install this beta. You are replacing IE 6. Assuming nothing goes horribly wrong, you should be able to uninstall it if necessary. But this is a beta version and you never know.

Update March 27: The second public preview was released recently and all the specific page layout issues I mentioned above are solved. But I did notice another issue with MediaWiki, the login bar at the top of the page is not placed correctly, normally it is on the right, with IE7 it is on the left under the site logo. Since this release is supposed to be layout complete, I guess MediaWiki developers will have to work around this minor problem.