Re: general design issue

Bob Badour wrote:
...
>>
>> Here are my results, using statements that might be closer to what you
>> want:
>>
>> select distinct substr(key,5,1) from words where key GLOB 'inte*'
>> had an ET of 9
>>
>> returned the letters g,l,m,n,r,s
>>
>> whereas
>>
>> select distinct substr(key,5,1) from words where key like 'inte%' had
>> an ET of 314
>>
>> So, about a 30:1 difference between LIKE and GLOB.
>>
>> BTW, this list has 331 words that start with the sequence 'inte'.
>
> In either case, did it take longer than a second to return the results?

I'm guessing these ET's are in milliseconds. I have no idea how memory was used and I put the words table on an sdhc flash card. If skirk is using an embedded device, probably my cpu is faster, I think it is an Intel Atom somewhere around 1.6GHz.

There does seem to be variation in sqlite's behaviour from my smaller test yesterday where I had problems with 'distinct' and 'substr'. Maybe the difference is because I didn't explictly build an index, just specified 'primary' for the 'key' column. Received on Fri Apr 24 2009 - 18:05:29 CEST