When using the <!--more--> tag to split your posts wordpress will generate a link that will take to a bookmark in the full post text.
I wanted the link to take the user to the top of the page so that the user will also see the page's header.
There is an entry on wordpress.org faq about this named "How can I customize the “more” tag?" that suggests a modification to: wp-includes/template-functions-post.php or wp-includes/post-template.php for wp 2.1 . The modification is fairly easy to do and it's about removing the "#more-id" anchor from the link but I don't want to do this because this means I'll have to remember to do it every time I upgrade wp.
My solution was to implement this as a plugin. Basically my plugin just removes the anchor from the link that takes you to the post page. This is slower then the modification suggested on wordpress.org but it's more convenient to me.
If you use authentication in squid you have several mechanisms ( authenticators ) to chose from. The Basic authenticator is the easiest to set up and the most insecure because the client sends the username and password in plain text to the proxy server.
Instead of using the basic you would consider using the digest authenticator. This authenticator does not require the client to send the user and password in plain text but encoded in an MD5 hash so that an attacker that captures the data between the client and proxy server will not be able to use the user and password.
You will need Wordpress 2.0.x, Google sitemap 3.0beta and Jerome's keywords 1.9 for this plugin to work. It would be easy to make this work with Google sitemap 2.x so if you need it just let me know.I don't know how this would work on older versions of Jerome's Keywords cause I haven't tried that.
The development version of FreeBSD ( 7.0 ) seems to scale a lot better then linux on SMP systems.
A combination of latest freebsd scheduler ULE 2.0 that is built into FreeBSD 7.0, the libthr threading library and a patch (not commited, yet ? ) that addresses poor scalability of file descriptor locking and some other patches is what made a system with 8 core amd64 cpu and 16 GB of ram outperform linux by a factor of 4 in MySQL tests.
The tests were performed using sysbench OLTP benchmark, a tool designed for testing mysql performance.
Brief test results: Linux is actually a little ( ~ 2% ) faster then FreeBSD for less then 9 clients, but when the number of clients grows to 20 linux performance drops a lot and FreeBSD's performance stays the same.
This is a maintenance release of the 8.0 code that fixes a few bugs as the original accounce says:
Tara! The first maintenance release of the 8.0 code:
8.0.1 (api:86/proto:86)
--------
* Fixed some race conditions that could trigger an OOPS when the loca disk fails and DRBD detaches itself from the failing disk.
* Added a missing call to drbd_try_outdate_peer().
* LVM's LVs expose ambiguous queue settings. When a RAID-0 (md) PV is used the present a segment size of 64k but at the same time allow only 8 sectors. Fixed DRBD to deal with that fact corretly.
