Entries Tagged 'Site related' ↓

I really like Dreamhost web hosting. Featurewise they provide more than any other shared hosting provider I’ve come across, and the freedom you get to install your own programs (trac, PHP, SpamAssassin, Django & Rails) is something I’ve only previously seen on dedicated servers or VPS web hosting (not that I’ve done many of those things, but hey - it’s nice to know it’s possible!).

In recent weeks Dreamhost has been having some problems. Actually that’s an understatement. They’ve been having big problems. Not as bad as the problems my last hosting company had which drove me to leave (at least Dreamhost always have backups to restore), but enough to make me question hosting commercial sites with them.

Is there an alternative to Dreamhost web hosting, that matches the features and freedom available? I’m happy to pay more… although obviously the cheaper the better! I cannot justify the cost of a dedicated server (it’s also too much hassle) and I’ve been put off Virtual Private Servers because the RAM isn’t burstable (I can see myself running out of this fast - many databases and concurrent visitors).

Some of the freedom is unnecesary: if Dreamhost configured their SpamAssassin carefully there’d be no need to run my own instance, so I can live without that.

Some other hosting company must have seen the need for geeky/techie hosting - if not there’s a (demanding) market out there!

I’ve always liked the idea of having related entries shown - it’s a smart system of navigation and helps explore the knowledge available. Hence the first plugin I installed on this Wordpress-based blog was Related Posts by Alexander Malov & Mike Lu.

The aim behind the related posts plugin is to show similar entries already made on your blog, regardless of their category. Think of it as displaying search results before the reader has searched. That’s how useful it can be if it works.

However, the Wordpress plugin did not give very good results when used on this site. Upon investigating the code, I discovered it was only using the post’s URL to compare the current post with other entries. Since my weblog URLs are based on the page title, which on this site doesn’t always highlight keywords from the entry, this was unsurprisingly not proving effective.

The plugin provided the option for you to put hidden keywords in your post which would be used. However I didn’t want to go back and add keywords to my old posts. Not to mention I’d forget the standard keywords between writing posts Tagging is a job for the computer, not a human…

So I’ve hacked up a modified version, which takes the post content, and calculates word frequency using weightings that can easily be modified (lines 57-62 of the code) for different parts of the post (eg URL, title, content). It then uses these details to match against the MySQL full-text index, (hopefully) returning more relevant results.

The script currently does not use any stemming algorithm or match against category IDs. Those are the most obvious features to add but the latter would require the MySQL query being reworked, and it was bedtime!

If you want to see the MySQL full-text ranking (as shown on this site next to each related posts link at present on this blog) to give you an idea of the certainty of each match being shown, uncomment line 161 in related-posts.php.

Download related-posts.php

The code is not supported and was hacked because I was curious about recommendation algorithms. Improvements or suggestions are welcome. You are free to use and modify this code.
Question: Is there some way to manage file uploads like above from within Wordpress? I’ve uploaded it manually as the ‘upload’ box on the “Write Post” screen said a php file wasn’t a valid image!

Finally I have a blog I can write to again.

The reason for the delay was changing hosting company (the subject of a future post) and moving to one that didn’t support PostgreSQL. I was running an ancient version of MovableType and didn’t have the MySQL schema available, and when I contacted Six Apart they said they couldn’t help.

Reverse engineering a database schema wasn’t my idea of fun, and since the Movable Type version was very old I decided to move. Not being able to make up my mind I have installed Wordpress, despite my previous reservations (and current ones - do these guys know anything about code architecture - the number of includes to dig through from index.php to find any ‘working code’ - or commenting code?) which is working, and an extra PHP file to handle rewriting all the old URLs.

So apologies to any of you who have tried commenting or searching the weblog in the past 3 months - it’s back working now and I will change from this default Wordpress theme once I get my head round things!

Seeing the logo over on the Drupal site started this off.

Oh pleeeaaaase. What’s got into everyone this year? The festival at this time is known as Christmas, therefore wishing Happy Christmas is correct. Countries such as Oman who don’t celebrate Christmas don’t have time off at this year, so Happy Holidays rubs their nose into the dirt even more.

Now assuming the fuss is because it’s a (supposedly) religious festival, if we’re going to call this Happy Holidays are members of other religions going to relabel their main festival to also be called Happy Holidays? Because you know I could get offended by seeing Happy Chanakuh or any islamic greeting, so they’d better change! Should I end up living in Saudi Arabia I hope they take note - they’d better not offend the minorities!

Perhaps more so, the U.S. festival of Thanksgiving should be changed so that one country’s holiday doesn’t offend everyone else. Being wished Happy Thanksgiving on mailing lists is so annoying; it’s not something I’ve ever celebrated or ever intend to celebrate. So rename it “Happy Thursday” and stop offending the rest of the world.

And how about the U.S. celebration of Independence Day? We don’t jubilantly mark winning the second world war every year or Nelson beating France, so cut your egos and stop celebrating beating the British. Grow up OK?

Thank you, and have a Happy Christmas!

Well, the “spam protection implemented”:http://peter.mapledesign.co.uk/weblog/archives/basic_comment_spam_protection.html on this weblog seemed to fox a lot of people yesterday Sorry guys, didn’t realise that the HTTP-Basic-Auth message wasn’t particularly visible in some browsers. I’ve added the username/password to the comment form; hopefully that should make it easier while still keeping spambots at bay.

I’ll be posting a roundup of weblog entries etc on the topic at some point - although probably not tomorrow! If I don’t post again, have a Happy Christmas, and try to separate yourself from the computer for a few hours…

I downloaded my emails this morning to find 89 trackback notifications sitting in my mailbox. One very simple database query later, and the obscenities are gone.

I’m pleased this is an important enough weblog to deserve so much attention, but if it happens again it will start to get frustrating

As you’ve probably noticed, the blogmarks are visible on the right menu (throughout the site - should be only on the index page, but I haven’t figured out how to do that in a MT global template). If you’ve not got Javascript enabled you won’t see them - the only way I could get dynamic content into a static MT-generated site. Archives will follow.

_[Apologies to Simon for nicking the format - well, they do say imitation is the sincerest form of flattery!]_
Continue reading →

Some things you never consider before you do them. Naming this blog “Perfectionism”, and the kind of search results that would attract was one of them.

So far this month over 30% of the search engine referrals were for the search “Perfectionism”. Last month it was 21.4% (335 searches). Now I’d like to think that these people remembered hearing about an excellent blog called “Perfectionism” and wanted to find it, but somehow I don’t think that’s the case

After over 2 weeks and not a single comment spam, I declare my “comment spam protection”:http://peter.mapledesign.co.uk/weblog/archives/basic_comment_spam_protection.html a success. Is anyone else using this method?

Having less Internet access at the moment than usual, and with an increase in comment spam on this blog, I’ve implemented basic comment spam protection, to prevent bots from posting comments. It’s just using .htaccess/.htpasswd protection for mt-comments.cgi. Hopefully it’ll work
If anyone’s interested, here’s the .htaccess file I put in my cgi-bin/mt directory (I tried putting it in the web root directory but it didn’t work - anyone know why?)

bc. AuthName “To prevent comment spam, please authenticate using: Username: **** Password: ****”
AuthType Basic
AuthUserFile /web/reywob/.htpasswd
Allow From All

require valid-user

I’d like to improve it to have a different/random username/password each time, as this would mean the bots would have to “learn” it every time; however, I can’t see a 2-minute way to implement this; and besides, it would be a pain while posting comments
h3. Update

Well, it’s been running for 4 days now, and no comment spam has appeared, which is a great improvement on before. Therefore I conclude that it is effective, subject to long-term distributed studies being carried out. _(can you tell I’ve had too many reports to write??)_