request for review

[Sebastian Kügler]

Hey,

One potential issue I see is that you ship the youtube logo inside the runner. 
We might not be allowed to redistribute it as that.

It also scales badly, so looks visually tarred.

Can you check the license of the youtube logo, and if we're allowed to ship 
it, make it look correct?

Also, enabling these runners by default, which you've essentially done, brings 
a great privacy problem -- everything I type into KRunner is sent to Google 
and Bing -- that's not acceptable as such, at least should be discussed before 
we introduce this kind of change. I was pretty surprised when I saw youtube 
movies in my krunner results. (Positive, because it works, negative, because I 
might have been looking for my furry porn collection, not something I'd want 
to tell any web services.)

Cheers,
-- sebas

On Friday, March 16, 2012 13:10:54 Shaun Reich wrote:
> i recently merged some new runners into master kdeplasma-addons
> without a review and i apologize for that (the motivation actually was
> because i was having some difficulties making some packages and had a
> few repos that needed it, but that's moot now).
> 
> either way, i've got the tar if you've got the feathers ;-)
> 
> i can file a formal review request if that makes things easier. i will
> do so later for the non-merged duckduckgo runner, so as to not repeat
> the same mistake.
> 
> but the only things of interest here are the youtube and bing runners.
> the bing one is the same exact thing, except for bing obviously.
> 
> https://projects.kde.org/projects/kde/kdeplasma-addons/repository/revisions/
> master/entry/runners/youtube/youtube.cpp
> 
> https://projects.kde.org/projects/kde/kdeplasma-addons/repository/revisions/
> master/entry/runners/youtube/tubejob.cpp
> 
> the code's relatively straight forward, because json > xml ;)
> 
> the reason why i had to use bing for image searching is because google
> is evil and their api has some silly 1k queries/day/key limit, which
> we could easily outcap. and ironically, bing doesn't have any limits
> on their api. i'm probably going to work on a flickr runner as well,
> because we need more integration with the internets.
> 
> i do have a question however, as to why matches get triggered (even
> for the mediawiki engine which currently resides in there, afict),
> even though it isn't in singlerunnermode, and the prefix isn't being
> hit. e.g. user is typing in "some query" would trigger a match for
> that runner, instead of it only being triggered by "wiki some query".
> any ideas on that?
> 
> seems like a waste of resources if it is in fact unnecessary.
> 
> there is an issue with the icon being installed though. i copied the
> icons from some kipi plugin, because they also happened to have a
> youtube icon set so that saved time.
> 
> however, it now means we have 2 sets of icons being installed. also,
> i'm installing the icons simply by running: kde4_install_icons(
> ${ICON_INSTALL_DIR} )
> 
> is there a good fix for this?
-- 
sebas

http://www.kde.org | http://vizZzion.org | GPG Key ID: 9119 0EF9