Optimisations

[pjrobertson]

Most of the information is in each commit message.

The reason I decided to GZip the requests to QSApp is to reduce bandwidth usage.
We can save quite a bit, for example on the plugins info .xml (from here)

http://qs0.qsapp.com/plugins/info.php is gzipped
Original Size: 218.07 KB
Compressed Size: 41.06 KB
Data Savings: 81.17%

GZip support isn't documented, bit I've used it fine in iOS.

The other changes:

• Better getter/setter methods, hopefully to avoid some crashes (see #1331 for an explanation)
• Fix this problem
• Reduce the number of calls to details, after Rob and I discussed it on IRC
• Pops history off the stack when browsing ←. Again, discussed on IRC. You'll notice at the moment if you dance up/down in a folder with →,←,→,← it doesn't work indefinitely. My changes fix this.

Lots of these commits are from a stash made a while back, and 'optimisations' can always be too optimistic, so it's worth keeping a close eye when running with this pull

[skurfer]

Looks like good stuff. One small thing I've seen in brief testing is that the action icon goes dim after a few seconds. Going to the second pane and hitting ↓ makes it look "normal" again.

If you're bored, here are some other small fixes we could add to this:

• The count at the bottom of the results doesn't always update correctly. Actually, I can only reproduce this by searching for "in" or "ni". The count is updated when I type "i", or if I type another letter after "in", like "ind". Weird. I may have to debug locally if you're not seeing this.
• If the last thing you had up was SearchEngine ⇥ Search For… ⇥ term, when you reinvoke QS, the "term" string in the third pane appears as the details, and the usual label area is blank.

[skurfer]

The count at the bottom of the results doesn't always update correctly.

Figured this one out. The status string is only updated if the selection changes. We might need to define a method to generate that string and call it for both selection changes and array changes (by overriding the setter for currentResults).

[skurfer]

I'll fix the two things I mentioned in another branch. Are you going to do as @tiennou suggested:

Change all accessors that use meta to go through setObject:forMeta unconditionally

I haven't looked to see if there are any places where that needs to be done.

[pjrobertson]

I checked through them yesterday, but couldn't see any extra calls to setObject:forMeta:, but I of course forgot to look for calls to [meta setObject:forKey:].

Another commit with a few more changes.

There are 1 or 2 methods which I haven't touched (setIdentifier and setName) as they do more funky things that I think we should leave alone