[Qt] tweak new peers tab in console window #4285

• remove starting height as table header and replace with ping time
• remove columnResizingFixer
• add local address (if available) in detailed node view (on top of the right view below the remote address)
• remove some .c_str() by using QString::fromStdString()
• rename Address to Address/Hostname
• rename secs to just s for ping time
• use MODEL_UPDATE_DELAY from guiconstants.h for the peer refresh time
• make PeerTableModel::columnCount() return no hard-coded value
• remove and cleanup dup private: section in RPCConsole header
• add new defaults for column sizes
• remove behaviour which keeps disconnected peers selected and also remove code which keeps track of last selected peer stats
• add sync height to detail view
• add some additional NULL pointer checks for clientModel in rpcconsole.cpp

@diapolo I suggest just cutting off long addresses (ie with ellipsis), then show the full address on hover. That's the accepted way of doing this. I remember there is some Qt mode for doing this, "elide mode" or something like that, but I'm not sure this is supported in tables. Edit: so no automatic resizing should happen based on the table contents!

@laanwj Small question, starting height means blocks a peer had when connecting to him or is that the current number of blocks it has? I think it would be better to replace starting height with something more valuable for a normal user like number of blocks on connect for example.

I also would like to re-order the information in the node details to be better or easier to read.

After #2784, I would suggest replacing starting height by pingtime.

Yes, ping time is nicer and more useful. @Diapolo It's the number of blocks on connect.

@sipa I'm fine with that change as it makes a whole lot of sense, when we auto-ping :).

@Diapolo #2784 has been merged

Yeah, I'm going to update this and include ping rather than starting height now :).

@laanwj You can check it out, before looking in detail at the code you should try and see if you like it. Perhaps you are able to add a better handling of column sizes to this. It looks okay with default RPCConsole window size, but doesn't scale or resize when changing the window size at all.

I used a Qt5.2 setting in the Qt Designer, which I now removed.

• The column for user agent is very narrow here by default. Handling of table sizing doesn't look like an improvement over what was there.
• I think the information pane is too wide now; bytes received/sent below each other was fine with me.
• The information pane is resizing in width all the time to the values, which looks very funky. The sizes should be fixed (certainly in width) IMO.
• Ping time should probably be in ms, as people expect that from other ping tools.

Hm the 'Node detail' problem also exists without this patch. The information pane, and the debug window around it, really shouldn't be actively resizing based on changing contents. I think this is a matter of setting some flag on the layout or widget.

• add sync height to detail view
• format ping time in overview table and detail view the same

@diapolo can we split this up a bit; all kind of changes all over the place are very hard to review and test

@laanwj Making further changes I can do seperate commits, which at the end can be just squashed. But I don't have any before the current state of this pull :-/. IMHO this changes a lot of things and should perhaps be considered the new base for the peers tab. Only real blocker is the layout problem IMHO, which likely will need your help.

My main gripe is still that this makes the peer information too wide.

Especially with a wide User Agent, it looks strange (the width of the user agents determines how far Sent/Received numbers are separated from each other).

I really preferred it as one-column layout (well - two column, label and information). If you change that back, this is ok with me.

@laanwj See the last squashme commit, this makes Send/Receive one line each again and adds a check if the nodeStateStats fetch was successful.

Looks good to me now. ACK after squashing.

Rebased into a single commit.

Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/p4285_a5b2d9c82e985fa8df4457072b4fa30781f1d53a/ for binaries and test log. This test script verifies pulls every time they are updated. It, however, dies sometimes and fails to test properly. If you are waiting on a test, please check timestamps to verify that the test.log is moving at http://jenkins.bluematt.me/pull-tester/current/ Contact BlueMatt on freenode if something looks broken.

Also I still think ping times would be better in ms instead of s. But that can be done later.

Sorry, I didn't catch that comment... see #4629.