ScientiaMobile

Hi Steve & Luca

I'm working on some regression tests, using the code formerly known as TeraWURFL, and seeing a bit of odd behaviour wrt. caching.

If I empty the cache table, and try to recognise this useragent;

Mozilla/5.0 (Linux; U; Android 2.2.2; ko-KR; Dell Streak)

I get the device ID: dell_streak_ver1

This is fine, but the data that is stored in the cache table looks like this;

mysql> select * from TeraWurflCache\G
*************************** 1. row ***************************
user_agent: Mozilla/5.0 (Linux; U; Android 2.2; xx-xx; Dell Streak)
cache_data: a:34:{s:2:"id";s:18:"dell_streak_7_ver1";s: ...

As you can see, the device ID in the serialised data is dell_streak_7_ver1.

So, any subsequent checks on that useragent will return the dell_streak_7_ver1 device ID, not dell_streak_ver1

Of course, this is not a huge problem, since both IDs are basically referring to the same device. But, it is a bit odd if the code is recognising a device as one thing, but then caching the recognised data as something else.

Good luck with the new venture.

David

Hi David,

This is very strange indeed! I'm not sure how this is happening, but I'll certainly look into it. A preliminary check on a few of my test boxes, including Tera-WURFL Explorer shows this device being detected as dell_streak_7_ver1.

Can you verify that the script you are using to see the ID is not actually showing the fall_back id? That would explain the behavior you're seeing since dell_streak_ver1 is the first fall back for dell_streak_7_ver1.

Hi Steve

Weird - it's not doing it anymore - I'm getting that useragent recognised as dell_streak_7_ver1 now, with an empty cache table.

The only change I made that could be relevant is to update to the latest wurfl.xml, which I did after I made my original post.

Sorry about that - false alarm. Still, at least I can remove my workaround from my regression tests.

All the best

David

Ok, well I'm glad it's resolved as it sounds like a very tricky problem to track down!