Re: [nattable-dev] Developer meeting

[Edwin Park <esp@xxxxxxxxxx>]

Hi Stephan,

This is great stuff! Yes we are definitely thinking in the same directions. Can you join our developer meeting tomorrow? Or let us know when you would be available to meet.

As for the current state of the experimental stuff, you can check it out here: http://git.eclipse.org/c/nattable/org.eclipse.nebula.widgets.nattable.experimental.git/

Much of it is written using Xtend at the moment, as I find it a lot more productive to work in than Java. The core layer api has been stripped down to its bare essentials: Layer, Axis, and Cell. There is a SWT and JavaFX renderer. Pixel locations are now in terms of doubles so that subpixel locations can be supported. The core layer api uses ints for column/row positions, but there is large layer support using BigInteger positions and BigDecimal pixel locations which are bridged by the ViewportLayer. There are also implementations for HideShowAxis, ReorderAxis, and CompositeLayer. There is still a lot left to be done: we have not added style configuration support yet as we want to make it compatible with CSS, event and command handling mechanisms are just barely stubbed out, no selection or editing yet, etc. It will be best to show you how everything works through a screenshare though, so hopefully we can get together and do that soon.

Best,

Edwin

On Wed, Jun 26, 2013 at 3:44 AM, Stephan Wahlbrink <sw@xxxxxxxxxxxx> wrote:

Hi Edwin, hi fellows,

Edwin Park wrote [2013-06-12 00:13]:

Fellow committers,

Dirk and I have set up a regular meeting every Thursday at 1:00pm GMT
(9:00 am in NYC, 2:00pm in London, 3:00pm in Germany) where we will be
discussing NatTable development. Please let either of us know if you'd
like to join in and we can send you the Google Hangout invite. You're
welcome to join whenever you can, however frequently or infrequently you
are available.

We'll be discussing ongoing work on NatTable 1.x at the meeting, but we
are also turning our attention to building out a next generation
architecture for NatTable. This involves several radical new concepts
including multiple renderers (we're looking at JavaFX and SWT), major
refactoring and simplification of the layer interface, CSS support,
support for arbitrarily large data sets, multiple viewports, nested
tables, etc. We have started building out some of this in the
experimental NatTable repo in git. If you'd like to see where we are
with it, please join the developer meeting!

unfortunately I could not yet participate in the meetings. Nevertheless here is some input.

It seems we had similar idea for simplifying the layer interface: make most API independent of the orientation. You called it Axis, I named it Dim:
https://github.com/walware/ecommons-uimisc/blob/1086ea90635821bfe3f82383133cf880bd914de8/de.walware.ecommons.inatable.core/src/org/eclipse/nebula/widgets/nattable/layer/ILayerDim.java .
This can really safe a lot of code, avoiding to write always the same stuff for columns and rows. Old implementations of layers have simple dim objects which forward the calls to the old API. Based on the new "dim" API I reimplemented for example the viewport layer and composite layer (I had to solve other issues). Of course, writing code for both orientations once is not limited to layers, the pattern can be also used for commands, events etc. The class AbstractSelectDimPositionsCommand/ViewportSelectDimPositionsCommand is an example for that in my branch.

Another major change I made is the support of large tables, see also:
http://www.walware.de/?page=/it/statet/log.html&jump=long_vectors .
I switch from 'int' to 'long' for positions, indexes and pixels. Among others, I also had to replace list and arrays with single position items by list with position ranges e.g. in the selection model/layer.

What are your plans for the next version? Will it use xtend? Please share your results of the meetings.

Regards
Stephan

Thanks,
Edwin