1. download the Cabal source package e.g. from HackageDB
2. unpack it in a directory
3. locate the directory in the console
4. use “cabal install –reinstall” and watch the compiling process

For the network-2.2.1.3 package I couldn’t use this approach because I always got the message that “getnameinfo” couldn’t be found.

I discovered this mail that lead me to the solution, but was kinda vague in the second step. This is why I’m publishing this post. Here are the 4 steps with a detailed second step:

1. download the source from HackageDB
2. modify configure.ac by replacing the first paragraph with the second one:
   dnl Under mingw, we may need to set WINVER to 0x0501 to expose getaddrinfo. if test "x$ac_have_getaddrinfo" = x; then old_CFLAGS="$CFLAGS" if test "z$ac_cv_lib_ws2_32__head_libws2_32_a" = zyes; then CFLAGS="-DWINVER=0x0501 $CFLAGS" AC_MSG_CHECKING(for getaddrinfo if WINVER is 0x0501) AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[$ac_includes_default int testme(){ getaddrinfo; }]])],[AC_DEFINE([HAVE_GETADDRINFO], [1], [Define to 1 if you have the `getaddrinfo' function.]) AC_DEFINE([NEED_WINVER_XP], [1], [Define to 1 if the `getaddrinfo' function needs WINVER set.]) EXTRA_CPPFLAGS="-DWINVER=0x0501 $EXTRA_CPPFLAGS"; AC_MSG_RESULT(yes)],[CFLAGS="$old_CFLAGS"; AC_MSG_RESULT(no)]) fi fi
dnl Under mingw, we may need to set WINVER to 0x0501 to expose getaddrinfo. if test "x$ac_have_getaddrinfo" = x; then old_CFLAGS="$CFLAGS" if test "z$ac_cv_lib_ws2_32__head_libws2_32_a" = zyes; then dnl CFLAGS="-DWINVER=0x0501 $CFLAGS" dnl AC_MSG_CHECKING(for getaddrinfo if WINVER is 0x0501) AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[$ac_includes_default int testme(){ getaddrinfo; }]])],[AC_DEFINE([HAVE_GETADDRINFO], [1], [Define to 1 if you have the `getaddrinfo' function.]) AC_DEFINE([NEED_WINVER_XP], [1], [Define to 1 if the `getaddrinfo' function needs WINVER set.]) EXTRA_CPPFLAGS="$EXTRA_CPPFLAGS"; AC_MSG_RESULT(yes)],[CFLAGS="$old_CFLAGS"; AC_MSG_RESULT(no)]) fi fi
3. run “autoreconf” in the source directory. If you don’t have autoreconf download msysDTK-1.0.1.exe (or a better version)
4. run “cabal install –reinstall”

I hope this will save a lost soul some time, one day!

I resist using CAPTCHAs whenever I can to offer my visitors the best user-experience. Who likes to comment a post when you additionally have to decipher symbols on an image only to type them into a textbox?

For this reason I activated Akismet on all of my blogs. Now it quietly checks every comment and removes it, if it is detected as spam.

Now it happens that I own a community site based on WordPress MU, enabeling every registered member to run their own blog.

Unfortunately Akismet needs an API-key which has to be inserted by hand on each and every blog. At least that’s what I thought until today.

There is an easier way to globally activate Akismet on WordPress MU and specify an API key that cannot be changed. Follow these steps to aktivate Akismet on all your WP MU blogs:

1. Move the “akismet” directory from the “plugin” directory to the “mu-plugins” directory,
2. move the file “akismet.php” to the “mu-plugins” directory,
3. open the file “akismet.php” with a text editor and add your API key on line 12.

Now you have activated Akismet on all blogs and the ability to change the key is gone.

With this setup users can’t deactivate the spam protection and you can apply yourself to more important things.
Like how to get more visitors to your community site.

Oh, and BTW: I accidently the whole site. But thanks to Google Cache I had a backup of all my posts. Just think about it when you accidental delete your whole database.

I assumed the mistake in my WSDL file, since I knew the function “foo” existed.

If you get the “Procedure ‘foo’ not present”-error, too you either:

• forgot to add the procedure “foo” to your SOAP server implementation, or
• you are using a cached version of your WSDL file at the server side.

I discovered that sometimes a cached WSDL file is used, although you are using this code:
ini_set("soap.wsdl_cache_enabled", "0");
If you aren’t sure whether you are using a cached version or not, you can either

• reboot the server to clean the cache, or
• delete all files in /tmp/ starting with “wsdl-” (rm /tmp/wsdl-*)

If not a cached WSDL file is causing the problem, you should check if you really have a function “foo” in your implementation.

I found the best practice debugging SOAP is using a log file. I use this code to catch exceptions in the server implementation:

$logfilePath = '/path/to/your/logfile.txt'; $debug = true; require_once('Zend/Log.php'); require_once('Zend/Log/Writer/Stream.php'); $writer = new Zend_Log_Writer_Stream(fopen($logfilePath, 'a', false)); $logger = new Zend_Log($writer); try { require_once('MySoapClass.php'); if ($debug) { ini_set("soap.wsdl_cache_enabled", "0"); } $server = new SoapServer($wsdl_url); $server->setClass('MySoapClass'); $server->handle(); } catch (Exception $exception) { if ($debug) { $logger->err($exception); } throw new SoapFault('MySoapServer', $exception); }

I’m using Zend_Log for the logging part, but you should get the idea.

For faster feedback I’m tailing the logfile (tail -f /path/to/your/logfile.txt) and use my usual SOAP client to do the requests.

This debugging technique is useful if you get errors like “looks like we got no XML document” on the client side.

How do you debug server-side SOAP code?

You create a famous wordpress plugin and implement a backdoor to inject your own links.

Josh, from Josh Team and Eli, from Blue Hat SEO did this with his WordPress plugin “PingCrawl“.

PingCrawl automatically searches for related posts on other blogs with Googles blogsearch and pings each of them to get a valid backlink. Awesome idea!

If you take a look at the code of the plugin you stumble over this function:
protected function injection() { $link = file_get_contents( 'http://www.bluehatseo.com/cgi-bin/findsites.cgi' ); if ( empty( $link ) ) { $link = '<a href="http://www.spottedhere.com">Dallas Nightlife Entertainment</a>'; } return $link; }

This function tries to get the content of the website http://www.bluehatseo.com/cgi-bin/findsites.cgi. If it doesn’t get one it will (throw an error and) return default linkcode to http://www.spottedhere.com.

If you read further you’ll notice this code, which injects the link at a probability of 1:6 1:4 for each tag:

$seed = rand( 0, 11 ); if ( $seed <= 2 ) { $content .= '<li>' . $this->injection() . '</li>'; }

You are save to remove these two code parts from the plugin, if you want to.

If you want to inject links with a sneaky plugin, you should probably not release it in a post on a SEO blog. Just my thoughts.

I hope Eli will not take this post personally, I just want to protect others who are too lazy to look over the code of PingCrawl.

Chris explains how he’s deploying web applications and why you should follow this six-step-approach, too:

1. The first deploy will not hurt, but as soon as you change something and it breaks it will hurt.
2. Make sure you are able to undo changes in the productive environment if you change something.
3. Don’t forget that there’s more in your application than code!
4. Automate every step, else mistakes will happen!
5. Make sure the deployment and production environment are as close as possible.
6. Ownership of code is important, make sure it’s properly deployed.

Chris recommends Capistrano, which is originally a deploy-tool for Ruby but since version 2 it doesn’t care about what it deploys. I personally haven’t used Capistrano, but hearing Chris’ satisfaction I think it’s worth checking it out.

I used Zend_Mail and Zend_Mime to create HTML-mails with inline images. I used the last stable Zend Framework version available right now: 1.0.3.

Microsoft Outlook 2007 displays odd characters, Thunderbird doesn’t

Check if you use:
new Zend_Mail('utf8');
If you do, change it to:
new Zend_Mail('utf-8');
Outlook doesn’t like the non-dashed version of the UTF-8 charset.

Inline images are displayed as attachments

You need to set the type of the mail to Multipart/Related:
$mail->setType(Zend_Mime::MULTIPART_RELATED);
Add the images like this:
$mail->createAttachment($imageData, $imageType, Zend_Mime::DISPOSITION_INLINE, Zend_Mime::ENCODING_BASE64);
Add the HTML-Version of the mail like this:
$mail->setBodyHtml($htmlMessage, null, Zend_Mime::MULTIPART_RELATED);

The subject displays odd characters

There’s a bug in Zend_Mail that causes subjects longer than 200 characters to be malformed in the mail. It’s also causing problems in the Zend Framework version 1.5 preview. The solution is to use subjects shorter than 200 characters until a fix is integrated in Zend_Mail.

Zend_Mail throws a 501 exception on international mail-addresses

If your email addresses are pointing to domains like hpphör.de, your SMTP-server might prevent you from sending to them, causing Zend_Mail to throw a 501 exception. A possible solution is to convert the domainname to the IDN equivalent (hpphör.de results in xn--hpphr-mua.de). Currently, there is no ZendFramework module available to do the conversion. Maybe the IDN Pecl package can help you on this subject.

Note: Malformed mail addresses also cause a 501 exception to be thrown. Catching a 501 exception does not guarantee an IDN domain-address!

The HTML layout is totally messed up

Don’t use CSS for styling, go back to Table-based styling. Most mail-clients still don’t display HTML correctly. Use inline-style definitions, even though they produce redundant code.

Get more information on HTML mail styling here:

• How to Code HTML Email Newsletters
• A Guide to CSS Support in Email

I hope this helps a bit if you find yourself in the position to code a PHP mail-script.

At SitePoint I found a very good article that covers how to design HTML mails from the content point of view. I recommend to read the first two pages if you need to write the content by yourself!