Posted on December 14, 2008 in BITW by adamNo Comments »

I found this installment of BITW while downloading the current issue of SD Times.



There are actually two bugs (at minimum) on display here

  • The most obvious is that there are two entries in this select box that do not have labels and so just display as ‘s. If I was troubleshooting this I would first look for incomplete records in the database and possibly suggest a ‘no null’ condition be applied to the column. If all the records that should be displayed are, then the likely culprit would seem to be a hardcoding of the records to display. (Or could actually be both.)
  • The other is a grammatical mistake (which if you have ever seen me on IM or IRC you will note the irony of me pointing this out). ‘s indicates possession. For example, Adam’s blog. Doing this sort of dynamic pluralization opens up a world of potential hurt as there are so many edge cases (such as SD Times’ bug). It also means that you cannot (easily) internationalize the site. Interestingly, the first item in the list is correct.
Posted on August 24, 2008 in BITW, Quality by adamNo Comments »

Different companies react differently (which makes sense) to Bug In The Wild reports. Too often you hear about reports in a security context where companies have responded with a pack of lawyers. EA however has recently illustrated how to capitalize brilliantly to a BITW.



I originally found this on Neat-o-rama where someone in the comments suggested that the original report might have been faked to setup the response, and I can certainly see that happening. But it is too good a concept to let that overshadow it.

Posted on June 19, 2008 in BITW, Quality by adamNo Comments »

I’m pretty sure that cruisecontrol.rb does not hand svn:externals properly so I’ve have to explore regexes in Ruby. In doing this I found the super handy Rubular online regex testing site. It is certainly easier to debug these things in realtime in a browser than kicking off a full cc build.

There is however a tiny glitch in it.

In HTML, multiple whitespaces in text get merged into a single one. If you want/need multiple ones then you need to insert the ‘not breaking space’ HTML entity code; & . The results of Rubular appear to be neglecting this.

Using the regex

(.*-r[ ]{2,}\d+\s+|\s?)(.*)

against the string

geokit -r       45 svn://rubyforge.org/var/svn/geokit/trunk

results in the following matches

  • geokit -r 45
  • svn://rubyforge.org/var/svn/geokit/trunk

Remember the rule about whitespace?

The first match should have been: geokit -r        45 

The devil, as is often the case, is in the details.

Posted on June 17, 2008 in BITW by adam1 Comment »

In what could be the best bug-in-the-wild ever, it appears that a recent patch to Age of Conan had the unintended consequence of changing how endowed the female avatars are. But players, mostly the teenage male demographic naturally, have been told by the publisher not worry as they “are working on a fix for this and your breasts should be back to normal soon.”

Some lessons:

  • Always keep an eye out for accidental breakage; even in seemingly unrelated (body) parts
  • Know your audience and what they want. That knowledge can help you tailor your release checklists

I’m not sure what is most entertaining about this bitw

  • The outcry
  • That it got missed
  • That “check breast size” is now on a checklist somewhere
  • A PR person came into work and had to spend the day wordsmithing responses to complaints about digital bust sizes

I can’t claim to have discovered this; the original article that passed by me is at Wired.

Posted on March 19, 2008 in BITW, Quality by adamNo Comments »

I’ll start posting these as I find them (jumping on the Bug In The Wild bandwagon started by Adam White (the other Toronto Adam tester)).

This one is from CafePress’ newsletter to shopkeepers about a sale that is going on right now for them. If you look at the bottom text you will notice something not quite right, specifically with around the email to ping in order to unsubscribe.



(Direct link to fullsize (more readable) image.)

Now that we can see what the bug is (a placeholder wasn’t replaced with a real email address), where is the bug?

  • In the page copy?
  • In the placeholder itself so it wasn’t replaced?
  • The tool that was supposed to replace the placeholder?
  • The process around which the page was created?