> On Jan. 17, 2011, 11:25 a.m., Andrea Diamantini wrote:
> > uhm... it seems faster, you're right. What I could not properly understand 
> > is why you are actually NOT saving a ResponseList and let rekonq parse it 
> > everytime. Am I missing something?
> 
> Lionel Chauvin wrote:
>     A Response is a structure. For save this structure, I must use a format. 
> What format is better than the original format ?
> 
> Andrea Diamantini wrote:
>     I see what you're saying. But I'm thinking about this: we have different 
> original formats, but just one we use for responses. Moreover, having cached 
> responses just parsed save one operation more (the parsing). Isn't this the 
> heavier operation here?
> 
> Lionel Chauvin wrote:
>     "we have different original formats, but just one we use for responses"
>     no
>     
>     We parse json or xml files and we save needed data in a list of 
> structures (ResponseList).
>     We don't have code for save a ResponseList in xml or json. This is why I 
> save the dowloaded files to disk (For being parsed again later).
>     I think it is not interesting to create a new format for save and load a 
> ResponseList.
>     
>     "Isn't this the heavier operation here?"
>     heavier ? CPU usage or time ?
>     
>     Download from the web take the most of time.
>     Now we prevent to download data again.
>     
>     Disk access is may be what take the most of time (more than parsing).
>     We can perhaps keep in RAM a limited number of ResponseList. For 
> instance, if you press backspace we load and parse the file again.
>     
>     I agree it can be optimized but it seems to be acceptable, even on my old 
> laptop.
>     
>     What I would work on is the timeout: when download instead of use the 
> cached data ?

You win :)
Merging it ASAP.


- Andrea


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://git.reviewboard.kde.org/r/100398/#review945
-----------------------------------------------------------


On Jan. 16, 2011, 4:01 p.m., Lionel Chauvin wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://git.reviewboard.kde.org/r/100398/
> -----------------------------------------------------------
> 
> (Updated Jan. 16, 2011, 4:01 p.m.)
> 
> 
> Review request for rekonq.
> 
> 
> Summary
> -------
> 
> it is the first step of a cache system for suggestions.
> keep downloaded suggestions in memory.
> prevent downloaded it again.
> the completion list is more reactive.
> 
> 
> Diffs
> -----
> 
>   src/opensearch/opensearchengine.h c05501f 
>   src/opensearch/opensearchengine.cpp ef0f3f7 
>   src/opensearch/opensearchmanager.cpp 41e46ec 
> 
> Diff: http://git.reviewboard.kde.org/r/100398/diff
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Lionel
> 
>

_______________________________________________
rekonq mailing list
[email protected]
https://mail.kde.org/mailman/listinfo/rekonq

Reply via email to