[GCompris-devel] Regarding the new approach for search feature

[Johnny Jazeix]

Hi,

Bruno answered in the main thread that we can  keep the actual version
as it is not really slow but I'll still try to answer to the questions
from my pov.

2016-01-29 8:11 GMT+01:00 RISHABH GUPTA <rishabh9511 at gmail.com>:
> Hello,
>
> After reading the latest comment from JohnnyJ here
> https://github.com/bdoin/GCompris-qt/pull/110
> which was regarding the explanation of the new approach that bdoin suggested
> ,I now understand the new approach a  bit better but would still need some
> time to think about how it could be coded.
>
> Thinking about how this could be coded is a secondary issue for me ,what's
> more important is that  we first discuss whether or not it will be  worth it
> to take out time for this new approach and replace it with the existing one
> ?

std::accumulate can help ;).

>
> I will list some points both from the  users and developers point of view
> that might help us take a decision on this new approach :
>
>
> User :
> 1.Do we get better search results with the new approach ?
> Moreover how exactly do we define that one search result is better than the
> other?To me it's displaying the activities in the ascending difficulty order
> ,which is good for the kids and which  we are already doing with the
> existing approach
>

The search will be more accurate if we name the activity/section
correctly -> the activities will be sorted by score (so the one which
match the most will be displayed first).

>
>
>
> Developers:
> 1.What do we gain with the new approach
>   a. less memory usage?

I don't think because we'll have to store the store.

>   b. less time for searching the activities?

I'm not sure because we'll have to compute the score for each activities.

>   c. better code readability ?

This depend on how it is coded but the principle is more difficult to
understand.
>
> If the answer to the above points is yes then how much do we gain ?
>
>
>
> We have already been working on the optimisation and code readability part
> of the existing feature and are almost done with that .So if we don't
> diverge our mind on thinking about this new approach ,in very short amount
> of time we will able to bring the search feature  to GCompris.
>
> Thank you :)
>

For me, as well, we can keep what you did, so when you are ready, feel
free to update the PR.

Johnny