WebTech

For a limited time only you can apply to work directly with me at TMTI Ltd as a web programmer. Here's the job advert:

Visit the TotalJobs website via this link to apply!

Position Filled!

AVG 8.0 has been released and I've duly upgraded to it from 7.5. Unfortunately the new “Web Shield” component in AVG 8 yields dire performance in the Opera browser (version 9.27). I was beginning to blame it on the browser, my ISP, some of the sites I use (Bloglines, BBC News and Scrabulous) or a problem with WinXP but suddenly remembered I'd upgraded AVG a couple of days before and indeed disabling the “Web Shield” feature in AVG returns Opera to its normal sprightly self.

Unfortunately turning off the feature shows an exclamation icon for AVG in the system tray which is rather annoying, so I've just hidden the AVG icon entirely through “Customize (sic) Notifications” in XP. Hiding the AVG icon is a bit annoying because I notice it not being there and so from time to time ponder whether AVG is running at all, though at least I don't have to look at the displeasingly gaudy new AVG icon.

Not to worry though, turning off “Web Shield” isn't much of a loss if you use Opera, using Opera in the first place is probably your best shield from web-based exploits amongst modern browsers.

The latest and greatest (so far) PairX simulator has just launched: motorola.pairx.co.uk. PairX is one of the projects I work on full-time for a living at TMTI Ltd as Development Team Leader. Shame though you can't see the rather sizeable and entirely AJAX-based CMS running under XULRunner that is used by our in-house team of Bluetooth researchers to populate the content

I don't know about you, well, only what your web browser says about you, but I find the BBC Newsonline breaking news ticker ironic. Ironic because it's mean't to be informing us of breaking news and yet does so so painfully slowly. It's also inaccessible, not particularly usable and poorly implemented.

As a web programmer if I'd been asked to implement this feature then I wouldn't have used an iframe, especially one that downloads an entirely new HTML document on every request of the parent page — a non-cacheable document that contains the script engine inline (so you download it afresh every time you browse to the front page) as well as the server-side generated inline news ticker data.

If I'd designed the news ticker then there wouldn't have been an iframe, rather the news ticker items would have been served in the main page and styled as an unordered list. Then, when the page loaded script would convert this plain old markup into a ticker. This simple step alone would make the breaking news stories highly accessible and save on publicly-funded hosting costs since the script engine would be cached for future use and the data would already be in the main page.

I'd also make the ticker more usable at the same time — it would scroll vertically displaying whole sentences or chunks of sentences at a time and if a user hovered over the ticker whilst it was doing its thing I wouldn't just carry on regardless but would immediately jump to the end of the scroll routine and display the sentence or sentence fragment the user was hovering over on the assumption that they were hovering over it because they actually wanted to read it, and then I'd hold there until they moused away. Additionally breaking news items that didn't have an article to link to (because the news team hadn't written one yet) wouldn't be rendered as links that simply reloaded the front page — these links only confuse the reader who expects to read an article when one of these links is clicked and not be sent to the front page without explanation.

With some of these thoughts in mind I wrote for my own personal use a bookmarklet that neuters the BBC Newsonline ticker and displays the breaking news stories as a plain old HTML list, sans scrolling. I'm now making it publically available, but bear in mind that it only works in Opera and Firefox and not Safari or IE6 (don't know about IE7).

To use the bookmarklet simply bookmark the following link (or drag it onto your bookmark toolbar / menu): BBC Unticker. Then when viewing BBC Newsonline simply click the new bookmark.

The cranky user: What ever happened to Web engineering? [via Slashdot]

For a few of weeks at least, possibly longer, I've been noticing hits purportedly from search results on Microsoft's LiveSearch website, e.g.: "http://search.live.com/results.aspx?q=SEARCHPHRASE&mrt=en-us&FORM=LIVSOP". Whilst it looks like legitimate traffic a little investigation raises some questions, from which you could deduce that the traffic is referrer spam unless there was a techincal explanation.

The referred LiveSearch web address contains the search term that would ordinarily have been used to generate the page of search results that are being referred. This information is typically extracted by statistical packages such as Awstats that process information about website visits from the webserver's log files. If the referrer hits were referrer spam they'd increase the importance of LiveSearch as a search engine as seen in website statistical packages. As to whether the traffic is legitimate is tricky, but a couple of things suggest that the traffic could be fake: the IP addresses of all of these hits are owned by Microsoft (not those of end-users). Secondly, the search terms are common single words that wouldn't return search results where my site was on the first page of results displayed in LiveSearch (regardless of how many items are displayed per page) — entire search phrases such as "search", "slightly", "pause", etc.

Initially I thought the traffic was bona fide referrer spam, but I now suspect that it is legitimate traffic. The reason I say this is because for users with javascript enabled LiveSearch overrides the click action of search result URIs, redirecting them to a “GLinkPing.aspx” page which may be constructing its own HTTP_REFERER after tracking which link the LiveSearch user has used (explains the Microsoft IP block). One reason the referrer search term is a single word could be because the generation of the pass-through referrer has a bug that only catches one word of the search phrase and loses the rest, so I get a referrer that doesn't represent the user's original search and explains why I don't appear as a link on the referrered page of results.

I have to come to this conclusion because the alternative seems unlikely (Microsoft having a farm of robots referrer spamming fake hits for their search engine for PR purposes) and I'm mindful of Hanlon's Razor!

Update

An example of odd LiveSearch hits from my server logfile viewer showing hits from Microsoft IPs with Live Search referrers. The “Search: ” part is an automatically prettified way of writing “http://search.live.com/results.aspx?q=slightly&mrt=en-us&FORM=LIVSOP” that pulls out the referring search-engine search terms.

Update
Note: Since the time of writing the number of these hits has increased, or at least have become more noticeable in my logs, though the inanity of the referrer search terms has not changed. This leads to the conclusion that it is not an innocent mistake. So much for “never attribute to malice that which is adequately explained by stupidity”.

Perhaps it's old age setting in, but of late I've been falling foul of Opera's “Block Content” filter. If you don't know what it is then it's probably similar to “adblock” in Firefox (though I'm only guessing here). To use it you bring up the context menu on any webpage, choose “Block Content”, then click on any of the advertisements or other content you don't want to see again, save your changes and that's the end of any annoying adverts.

However, you can use it to not only block inline content, but to block any content including entire domains. Whilst this is extremely handy (e.g. no more doubleclick or googlesyndication adverts) you can fall foul of it very easily and not realise.

For example, I recently submitted a “site problem” report to Opera because I couldn't get CNN to stream video yet it was fine in Firefox and IE. Given the history of CNN and cross-browser problems I assumed it was either CNN or Opera that was at fault. Turns out it was me because I was blocking video content that also shared a URI fragment with CNN's adverts.

Then a couple of weeks ago I filed another “site problem” report to Opera because the links in an email newsletter (Sky & Telescope magazine [S&T]) that were encoded and attached to a redirector script at the clickability.com website were not being redirected to the S&T website, they would just hang and not redirect. I even sent an email to the helpdesk at the clickability.com website describing the inconvenience of not being able to follow any links in newsletters that used their hit-tracking redirectors.

The CNN issue was fixed when I remembered about the blocked content feature (you forget about it as it runs silently in the background) and wrote a better rule for CNN. However, the clickability issue didn't make me think about blocked content until this morning. Sure enough, at some point in the past I'd added the clickability.com domain to the blocked list and removing it fixed the problem.

Of course some poor tech somewhere is probably wondering about a bug report that he or she can't reproduce. As clickability.com provided an email address I've apologised and briefly explained the situation, though as Opera's bug reporting system is closed to the public I can't add a note and close the bug report so that they don't need to invest time and money trying to fix a non-existent bug.

However, Opera do need to do something about the silence of the blocked content feature as it's not very usable at the moment. For example, the problem with clickability.com would have been apparent immediately had Opera displayed a notice that the page had been blocked, instead of doing absolutely nothing at all. The sidebar panel could have the option of showing what media, if any, had been blocked in the page because of content blocking. Opera have added many new features over the last couple of years, but they do seem to have dropped the ball on usability in the rush to get features out of the door.

In the event that Opera crashes, Windows spawns the following web page with some helpful recommendations that user's of non-Internet Explorer browsers might find a little ironic:

Helpful advice from Microsoft. Highlighting is mine.