mikerowecode

I’m visiting family in the USA for Christmas. You can keep up with what we’re doing (well, at least photos) on my family blog.

I’ll be resuming my regular cracking pace of blogging in the new year… :-)

I’ve just started using Apple’s Aperture to manage my photos (family and other), and I’m pretty pleased with it so far. One thing, however, that is harder than it needs to be is entering captions. As far as I can tell, the only way to do this in Aperture is to click on the photo in the browser, mouse over to the Information panel and click in the caption field (making sure you’ve selected IPTC fields to be displayed), type the caption and mouse back and click in the browser window so it has focus (otherwise any further keys you press will still be typing in the caption field). Then click on the next photo. Eeew.

This is bearable for one or two photos, but when you have a largish batch to do, it is very tedious. It’s also prone to error, such as when you start typing the caption without the field having focus, and Aperture gleefully trying to act on all your keypresses.

So I threw together a small AppleScript that provides a marginally friendlier interface. This is one of my first attempts at AppleScript, so I’d be happy to receive any feedback on better techniques.

I just got a new can of deodorant, and it says “shake well before use”. I figure the night before should be long enough. :-D

Last Saturday, I learnt something. (Not that this is an unusual occurrence, of course.) I was trying to upgrade my Ubuntu box to “edgy eft”, and was having trouble downloading packages. It turned out that my ADSL had gone down. And up. And down. sigh So, after a few reboot cycles of the modem and my firewall, I was back up again.

The upgrade completed without too much pain (nice one, Debian and Ubuntu), so I turned my attention to some web site work I needed to do. Only could get through to my web server. Or mail server. Great, I thought, I just get my ‘net connection back up and my hosting service goes down. Good excuse to watch the Grand Final, if nothing else. :)

Several hours later, football over for another year and my servers still not accessible, I thought I’d better look into it a bit more closely. My hosting service provides ssh access to the Xen console of my machines, so I tried that. And it worked. And what I saw was many kernel messages about packets from a “bogon” IP address being blocked. And the IP in question resolved to an iinet.net.au (my ADSL provider) domain name. Hmm.

After a bit of investigation, it turned out that with the outage in the morning, my ADSL IP had changed from a 203.214.. network to 124.168... And this netblock has until relatively recently been “reserved” by IANA. As it happens, Shorewall keeps a static list of “bogon” (i.e not normally seen on the public Internet) IP addresses in a configuration file, and this list included the rather sweeping netblock of 96.0.0.0/3, which results in everything from 96/8 up to 126/8 being blocked. But at the start of this year, IANA released the class As from 121/8 to 126/8 to APNIC, who gave them mostly to large Australian ISPs as far as I can tell, and evidently some those (like iiNet) are starting to roll them out to customers.

Anyway, a quick poke around with Google turned up this updated bogons file, so I installed that and restarted my firewall, and everything was fine again. It all goes to show, nothing is simple in the world of internetworking.

When I tried to send an email on my new iiNet last night, I discovered that outgoing port 25 was blocked. This is quite common with ISPs, since it tends to prevent a good deal of spam and other malware activity. But it’s a nuisance for me, since I like to send mail via my own mail server.

But, unlike many other ISPs (not looking at anyone cough bigpond in particular), iiNet let you turn off the blocking if you know what you’re doing.

Nice touch.

Summary pages in Drupal, such as this one, display “teasers” of the nodes (pages or blog entries) if the entry is longer than a certain (configurable) length. If the entry is trimmed this way, Drupal helpfully displays a “read more” link below the teaser. This is ok, but it can be easy to miss if you are just looking at the text of the article, and mentally “tuning out” the noise of the links below.

So I modified my theme’s node.tpl.php (part of the now-default PHPTemplate engine) to hack in a “more” link at the end of the last paragraph:

--- node.tpl.php        (revision 43)
+++ node.tpl.php        (working copy)
@@ -9,6 +9,11 @@
            <span class="taxonomy">[<?php print $terms?>]</span>
          <?php }; ?>
        <?php }; ?>
+    <?php if ($readmore) {
+        // insert a link before the closing P tag if there's more to read
+       $read_more = ' <span class="readmore">...<a href="' . $node_url . '">more</a>...</span>';
+       $content = preg_replace('/<\/p>$/', $read_more . '$1', $content); 
+    }; ?>
     <div class="content"><?php print $content?></div>
     <?php if ($links) { ?><div class="links">&raquo; <?php print $links?></div><?php }; ?>
   </div>

This requires a variable, $readmore, to be set in template.php:

function _phptemplate_variables($hook, $vars) {
    switch ($hook) {
    case 'node':
        $vars['readmore'] = ($vars['node']->teaser && $vars['node']->readmore);
        break;
    }
    return $vars;
}

(and some styling in style.css, but you can figure this out for yourself ;) )

And now I have discrete little “…more…” links at the end of the last paragraph of trimmed article summaries.

I kind of feel like this should have been possible by simply adding the link after $content rather than stuffing something into it with a regex, and causing it to be displayed inline with CSS, but my CSSfu is not strong enough. Suggestions welcome…

After a frustrating weekend of ADSL outages a couple of weeks ago, thanks either to my ISP or, more likely, their upstream provider (the Australian Monopoly Telco Who Shall Not Be Named), I started hunting around for a new ISP. Just on a whim, I checked my exchange on the ADSL2+ roll-out status page at iiNet. And lo! It’s available!

So I signed up for their standard plan, which gives me “up to” 24Mbps and 10GB download per month (plus another 10GB off peak). Given that I haven’t had to try too hard to keep within my current 5GB (+5GB) plan, and iiNet’s 10+10 is $20/mo less, it seemed like a good place to start. I also get their local phone service (bundled with ADSL2+) and VoIP, which should be interesting.

The downside of upgrading from ADSL to ADLS2+ is that there’s no “fast churn”. In other words, I had to disconnect my current service and then wait for the new one to be provisioned. But now it’s active! (And only three business days later, which is better than I expected.)

I don’t have the new ADSL2+ modem yet, but even with my current modem (a Netcomm NB1300) I’ve seen download speeds of about 380KBps (that’s bytes there sonny), which is fairly impressive. Uploads are going at about 60-80KBps, which is a lot better than the old 24KBps. My line attentuation is hovering around the 50dB mark according to the NB1300, which I’m led to believe means I probably won’t get much more than 10Mbps anyway, but we’ll see how it goes when the new modem arrives.

It’s amazing how long five days felt without ‘net connectivity, though… :-D

I’m giving up on blosxom. Not that there’s anything wrong with it, but the advantages of keeping my blog in Drupal are too great to ignore.

I liked the idea of blosxom, and I love having a local copy of my blog in plain text (and being able to manage it in subversion), but the hassle of maintaining another bunch of HTML/CSS, and the extra work involved in adding functionality like comments, is just not worth it.

So, look over here for future (frequent?) posts. :) I’ll get around to bringing the blosxom content across and putting in redirects soon.

I just spent a frustrating couple of hours beating my head against a Drupal site, trying to change the site name (that appears in the site banner and title). No matter what I set in the admin/settings page, the site name remained unchanged. The database was being updated correctly:

mysql> select * from variable where name = 'site_name'\G
*************************** 1. row ***************************
 name: site_name
value: s:20:"My Fancy Drupal Site";

but the old name was still displaying on the site (and every time I went back in to admin/settings, the name had reverted).

Finally, I remembered this section of settings.php:

/**
 * Variable overrides:
 *
 * To override specific entries in the 'variable' table for this site,
 * set them here. You usually don't need to use this feature. This is
 * useful in a configuration file for a vhost or directory, rather than
 * the default settings.php. Any configuration setting from the 'variable'
 * table can be given a new value.
 */
$conf = array(
  'site_name' => 'My Drupal Site',
//  'theme_default' => 'bluemarine',
  'anonymous' => 'Visitor'
);

D’oh!

So I commented out that 'site_name' line, and everything worked as expected.

This is a test of the patched weblogger.el I found on Friends of Bart. It would be really cool if it worked, because emacs is like a second home to me. :)

There’s a small bug in the just-updated Debian Drupal package in sarge (see bug #368835). If you just installed the updated deb from security.debian.org (4.5.3-6.1sarge1), you may see this message when trying to access your drupal site:

Parse error: parse error, unexpected T_STRING in /usr/share/drupal/includes/file.inc on line 106

This is caused by a missing semi-colon in file.inc (on line 106, surprisingly enough). This patch fixes it:

--- file.inc    2006-07-27 21:42:57.000000000 +1000
+++ file.inc.NEW    2006-07-27 21:42:51.000000000 +1000
@@ -102,7 +102,7 @@
       fclose($fp);
     }
     else {
-     $message = t("Security warning: Couldn't write .htaccess file. Please \
    create a .htaccess file in your %directory directory which contains the \
    following lines: <code>%htaccess</code>", array('%directory' => theme('placeholder', \
    $directory), '%htaccess' => '<br />'. str_replace("\n", '<br />', \
    check_plain($htaccess_lines))))
+    $message = t("Security warning: Couldn't write .htaccess file. Please \
    create a .htaccess file in your %directory directory which contains the \
    following lines: <code>%htaccess</code>", array('%directory' => $directory, \
    '%htaccess' => '<br />'. str_replace("\n", '<br />', \
    check_plain($htaccess_lines))));
       form_set_error($form_item, $message);
      watchdog('security', $message, WATCHDOG_ERROR);
     }

This appears to have already been reported in bug #380022.

I am playing with the “trunk” version of Django, a nifty python-based web framework. This evening, I spent a few hours getting back up to speed with the current state (trunk is changing fairly rapidly at the moment, all for the good though). But I kept getting TemplateSyntaxError in the admin module, with a message something like:

django.templatetags.adminmedia does not define a variable named "register"

(Sorry that’s not verbatim, I stupidly didn’t copy the error before I fixed it.)

I double checked all my settings and removed references to any of my local code, but kept getting the error. Finally, someone on the IRC channel (#django at irc.freenode.net) suggested cleaning out the django code to remove any *.pyc (compiled python) files.

Sure enough, doing this:

cd ~/src/django.svn/trunk # or where ever you checked out django 
find . -name "*pyc" -delete

fixed it.

Last year, I bought some EarJams for my iPod.

They were passable, but I was getting increasingly frustrated with the lack of accuracy or balanced frequency response. In fact, I reverted to using just the naked Apple-provided buds, since at least they were more or less faithful to the original sound (if a little tinny unless manually held firmly in ones ear).

So for my recent birthday, thanks to some cash from my mum and dad, I splashed on a pair of Shure E2c in-ear ‘phones. And it’s probably the best technology purchase I’ve made in a while. They feel “serious” (solid is probably not the best word to describe in-ear head phones), the cable is substantial without being too stiff, and they look pleasingly “techy”. But most of all, the sound.

A friend had told me about a pair of Etymotics he had tried, how he could hear all sorts of things in recordings he’d never noticed before… and it’s the same with these Shures. The level of detail I can now hear, especially the live sound of classical recordings, is incredible. Of course, with some artists, like Glenn Gould, this is not necessarily a good thing. ;)

The overall balance is great, and the sound is lively and crisp across all musical styles—from thumpathumpa bass to an orchestral sound stage. I’m a happy camper. :)

As recently chronicled, my Powerbook is no longer viable as a portable notebook computer. This makes me sad. :’(

The obvious silver lining, of course, is that it provided an opportunity for a new purchase. So I umm-ed and ahh-ed about the new MacBook Pro, but it was pretty hard to justify the four or five grand it would cost me. And everyone warns you off getting the first release of a new generation of Apple hardware.

Also, as much as I love the 15” wide screen, the idea of the 12” form factor was very attractive—most of the non-desk places I’m likely to use a laptop lend themselves to small and light.

So it came down to either a 12” PowerBook or iBook. I finally settled on the iBook. While it’s even slower than my old Powerbook, the price factor was too good to ignore. Essentially, it would have cost me about $1,000 more for the 20% speed boost and marginally smaller size/weight of the PowerBook. I can save that money now and upgrade to a high-end Intel model sooner rather than later.

And perhaps more to the point, the iBook is less “precious”… I’ll feel more comfortable throwing it in my bag and whacking it out on the table of the local coffee shop, not having to worry about sullying its lovely finish.

I’ll be keeping it away from open milk cartons though…

Well, some months have passed since my happy ending. Turns out it wasn’t the end, and it wasn’t so happy.

After a few weeks, my screen started playing up again. It was just an occasional flicker at first, but quickly degenerated into being almost unusable. I took the PowerBook back into the shop for another look, but they weren’t too optimistic. In a nutshell, they tried everything short of replacing parts, and the most likely part that would need to be replaced was the screen—at around AUD1,500, not really worth it.

The screen more or less behaves itself when left on for long periods (the problem seems to be partly heat related), so it was ok during the day at the office. But it is pretty useless for the train or coffee shop. So, I resigned myself to having a portable desktop. I bought a 17” LCD display for home, which along with an external keyboard and Mighty Mouse makes for quite a usable office machine.

Sigh. Maybe next time, I’ll get the extended warranty…

This morning, my mail server was rebooted by my ISP. This is not extraordinary in itself, but somehow it caused my mail client (Apple’s Mail.app) some confusion. It thought the IMAP server was offline (which it was for half an hour or so), and would not reconnect.

When I restarted Mail, it froze. “Application not responding”. Force quit worked to kill it, but it always froze on restart (even after a reboot). Other IMAP clients (I also use the excellent Mutt in conjunction with offlineimap, and when really desperate, Thunderbird) were working just fine.

I found this article on macosxhints, but its suggestion:

% cd ~/Library/Mail
% find . -name '.index.ready' -delete
% find . -name 'content_index' -delete
% find . -name 'table_of_contents' -delete

did not work. (Well, it obviously worked, the files were deleted, but it didn’t help the symptoms.)

I noticed another indexy cachey looking file in my mailbox directories: .mboxCache.plist. Deleting all of these:

find . -name '.mboxCache.plist' -delete

seemed to fix Mail.app for me.

NOTE I have no idea what this file is for, or what the implications of deleting it might be. It appears to have Worked For Me(tm), but it could also cause all your email to fall out of the back of your computer and make a nasty stain on your carpet.

I just tried to access one of my local dev sites (that is, running on my powerbook), and received the dreaded “mysql can’t connect to socket” error:

Can't connect to local MySQL server through socket '/var/mysql/mysql.sock'

I was sure MySQL was running (I could access it via the command line mysql tool, and MySQL Administrator). And it used to work… But sure enough, there is no directory /var/mysql, and no mysql.sock file in it. :-/

After a bit of digging around, I noticed that the php.ini file (/etc/php.ini.default) had been updated on Feb 8th, about the time OS X 10.4.5 was released. Hmm.

I don’t remember customising this file, but it certainly didn’t contain any customisation now. The “mysql.default_socket” setting was blank, meaning it would use the compiled-in default. I’m not sure whether PHP was upgraded with 10.4.5 and this changed the default, or if I had customised it previously and I’ve just forgotten. But either way, it was now looking for the MySQL socket in /var/mysql, and my version of MySQL (4.0.24) wasn’t putting it there.

So I created this dir and set its owner to mysql:

sudo mkdir /var/mysql
sudo chown mysql /var/mysql

and followed the instructions in this Apple support article by adding the following setting to /etc/my.cnf:

[mysqld]
socket = /var/mysql/mysql.sock

[mysql]
socket = /var/mysql/mysql.sock

(The last two lines are not mentioned in the Apple article, but are required for the command line mysql utilities to work.)

A quick restart of the mysql server later, and all is good again.

From Wil Shipley:

I never, ever fix a bug, even from another user, unless I can say with some surety that I can see the circumstances that led to that person writing that bug. It may be as simple as “oh, copy and paste error” or “this person was tired and overlooked this edge case, and that didn’t come up until now.” But if you see a bug and think, “How in the hell did this code ever work?” then you’ve got a lot more work to do (assuming the code did ever work), because you aren’t allowed to move on until you can answer that question.

Understanding your code, and the underlying code of the environment you’re in, is more important than fixing any one bug.

Sage advice.

If you’re using Firefox and are browsing a messy/ugly site that neglects to provide a “printable” version, you can turn off CSS styles. Just select View/Page Style/No Style from the menu. Often times, this will leave you with a nice plain page with Just The Content.

When I recently broke my finger (the pinky on my right hand) I realised that I’m a better typer than I thought. Because losing the use of just the finger has slowed me down significantly. I’m almost back to two-finger hunt-and-peck typing. Hopefully I’ll be able to un-learn the bad habits when it heals.

I suppose I should be greatful it wasn’t my left hand, since that would make using Emacs almost impossible.