[Search] Refactor: implement asynchronous search

[jku]

To prevent UI from blocking and to prevent excessive searches, the custom model needs to be refactored to not do synchronous searches in on_get_value().

This is achieved by

• adding SearchModel.set_request_range() so the view can inform the model of the required rows
• Adding a dict-based cache for the search results
• Adding row-inserted/changed/deleted signals to the model

The commit is fairly large, but I did not see a way to do this in parts and still keep the in-between commits functional.

Fixes #2

Signed-off-by: Jussi Kukkonen jussi.kukkonen@intel.com

[jku]

With regards to discussion in the corresponding media-service-demo bug:

• this commit does not use a ListStore but adds to the custom model that was already used: this is a bit more complex but is manageable and uses less memory than the ListStore approach in normal use even with massive media libraries
• the performance is quite good, better than with a ListStore for the pathological case: the UI is fine with even a million rows with no hiccups on scrolling. The only slowdown is when the (massive) TreeView is created, and even that could be optimized by attaching model to view after we know the model size.
• the case of servers not giving us max_items is not solved in this commit, but solving it is now possible: The Model now has a variable "may_have_more_results" which the view/controller could use to either show additional UI, or just try loading more items when the view gets closer to the end.

[jku]

This sounds fine to me. Although it ruins the commit author info, doesn't it? Is this why you guys use signed-off-by everywhere?

[markdryan]

Actually, we use signed-off everywhere because we are encouraged to do so by the 01.org guidelines.

https://01.org/community/participation-guidelines

And you are correct about the committer. If I squash two patches submitted by someone else, I am listed as the committer. The author is however still set to the original author. You can see this if you look in the commit logs in github. Squashed commits have both a committer and an author whereas as non squashed commits only have an author.

I hadn’t realised this before. Do you think this is a problem? If so perhaps we should stop doing this.

---

Intel Corporation SAS (French simplified joint stock company) Registered headquarters: "Les Montalets"- 2, rue de Paris, 92196 Meudon Cedex, France Registration Number: 302 456 199 R.C.S. NANTERRE Capital: 4,572,000 Euros

This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.

[markdryan]

You should update the Authors file as well.

[markdryan]

So Jussi, what should I do? Should I squash?

[jku]

On 6 March 2013 14:35, markdryan notifications@github.com wrote:

Actually, we use signed-off everywhere because we are encouraged to do so by the 01.org guidelines.

https://01.org/community/participation-guidelines

That makes a bit more sense -- I didn't know we had a "certificate of origin" and without it having Signed-off-by didn't make sense.

And you are correct about the committer. If I squash two patches submitted by someone else, I am listed as the committer. The author is however still set to the original author. You can see this if you look in the commit logs in github. Squashed commits have both a committer and an author whereas as non squashed commits only have an author.

I hadn’t realised this before. Do you think this is a problem? If so perhaps we should stop doing this.

Ah, that's true: the author of the squashed commit is the original author and not the "squasher". I didn't know that, that's very clever of git. I have no problems with the process in that case, I think it's just fine.

• Juss

[markdryan]

Squashed and pushed. Thanks.