At the start of this year, my friend and fellow tester Zeljko was working in Mexico, Guadalajara. You can read about his experience here. He provided a Picasa foto album, and, because he is a tester, Zeljko asked a question in this photo comment.
Here is my answer.
As I know that Željko has Croatian letter Ž at the start of his name, and that lettrer was printed as Ž, my first assumption was that Bikla application does not work with utf-8 encoding. First, I wanted to replicate the issue.
I first found one web page with letter Ž. UTF-8 hex code for letter Ž is c5bd (two bytes). In order to check has letter Ž been acctualy encoded in web page source using utf-8 encoding, save page source as html, open it in vim, switch vim to hex mode (:%!xxd), and search for c5bd. In right column, you will see text corresponding to letter Ž (but in vim hex mode it will be displayed as ..).
Other option is to check is your browser encoding set to utf-8 and if you can see letter Ž, then you know that letter has proper uft-8 hex code c5db.
In order to reproduce Bikla issue, after loading web page, I started to change my browser encoding. First in top-down list in Chromium is Western (ISO-8859-1). And browser changed Ž into Ž. Bingo, but this time I was lucky, but skilled tester would not guess, by try to use heuristics (more on heuristics you can find here) in order to determine Bikla page encoding from the first try (there is 37 encoding supported in chromium). I also confirmed that Zeljko’s name was properly encoded in the Bikla database. Question for Zeljko: how was your name entered into Bikla datbase?
Here is my heuristic. Go to google, google for ‘spanish character encoding’ and try those encoding. Google returned following encodings:
utf-8, iso-8859-1, iso-8859-e15.
So, this time, if I will go on the encoding list from left to right, I will get the Bikla encoding from the first try, but without any luck needed.
As I tester I identified following:
the issue: Ž into Ž
testing tools: vim, chromium, google
use of heuristic, how to reproduce issue
Knoweledge: character encodings.
Conclusion: I would inform Bikla that they should implement utf-8 encoding into their application. Their decision would be based on the fact does wrong printed names affects Bikla business.
For testers: find one more Croatian letter from the picture that was wrongly decoded.
Croatian character on trip to Guadalajara
Leave a Reply
You must be logged in to post a comment.
To answer your question about how my name got into the database: I am pretty sure I have registered here: http://bikla.net/index.php?Op=registrate&language=ing
If somebody from Bikla was entering my name, they would not have the letter Ž on the keyboard.
I am not sure how interested Bikla would be to fix the site. I was probably the only Non-Mexican in the history of the company to register. But, feel free to let them know. 🙂
So knowing that you entered in their registration form your name with Croatian letters (and that is what every tester should do), we know that Bikla database correctly stored your name. There are two reasons for that:
1. You browser encoding setting was set to utf-8
2. Bikla database was set with encoding utf-8.
3. business logic correctly transferred Croatian characters from presentation layer to database layer.
They only have issue with presentation/printing layer that does not correctly decode Croatian (or any character not from latin1 set) utf-8 characters.