Search against misc. dictionary info
- Thread starter mikeo
- Start date
The freeform text fields are part of it, but even searching on fixed tags like F.E. et al you still run into the issue of merging a couple of very big sets of search results; there are about 23,000 F.E. entries in the ABC, and on Palm OS you can't even reliably allocate a buffer big enough to hold all of those results at the same time.
Thanks, Mike, that helps to understand the problem.
To illustrate with my earlier example, the set of all xiao4, and the set of all FE, are both large sets. But the first set you already have, when you search for xiao4.
I'm wondering if it would be feasible to have a (probably very slow) process which ran against those results, one by one, marking only those xiao4s which also happened to contain an FE. The resulting group would probably be fairly small.
To illustrate with my earlier example, the set of all xiao4, and the set of all FE, are both large sets. But the first set you already have, when you search for xiao4.
I'm wondering if it would be feasible to have a (probably very slow) process which ran against those results, one by one, marking only those xiao4s which also happened to contain an FE. The resulting group would probably be fairly small.
It's certainly possible (very common CS problem in fact), but because you can't easily keep those sets in memory it does indeed get rather slow. There are ways of working around that (for example, creating an alternate version of the character index that was sorted the same way as the part-of-speech one and then going through them sequentially without having to search the entire results list each time), but they're somewhat involved programming-wise, and we're really trying very hard to avoid adding any more complexity like this until we've got a nice, stable, reliable system to do what we've already said we're doing in 2.0.