OpenLayers owes much of its functionality to sandbox innovations in the repository. At the FOSS4G conference in Sydney, I played a short animation representing commit activity in the OpenLayers trunk. As a way of acknowledging all of the great work done by sandbox developers, I wanted to post a similar animation that shows sandbox commits as well.
Here’s the video on Vimeo. Trunk commits in orange/red and sandbox commits in blue/green (with apologies to the color blind). Credits to code_swarm for the animations.
Several OpenLayers developers and volunteers gathered after the FOSS4G conference to participate in the Saturday code sprint. And guess what: a lot was accomplished! Here is a quick overview:
Julien also started looking closer at the Refresh strategy that Kris Geusebroek (Xebia) had originally contributed.
Robert Coup (Koordinates Ltd.) dug deep into memory leaks and came up with several patches, all of them with memory tests. He is still working on more leak fixes, based on an original patch from Kris.
Klokan Petr Přidal provided a small but effective patch to support tile grids with truncated tiles at the bottom and right grid edges. This is necessary to support layer types like Zoomify, for which another patch is awaiting review.
Marc Jansen (Terrestris) started cleaning up the examples: he added more keywords, made improvements to descriptions and made notes about deprecated practices that some of the examples still show.
Volker Mische (LisaSoft) put some efforts into making it possible to drag OpenLayers maps with the moving mouse cursor leaving the map viewport (like e.g. in Google Maps). There is still some work required, but I will help Volker and this fix should be in trunk soon. On a side note: this issue is very old - ticket #39
Roald De Wit (LisaSoft) made the pink tiles for unavailable image tiles configurable with CSS.
Roald and Volker also went to the drawing board with me, and by looking at the slider of the PanZoomBar control we found a way how a new control base class could separate the UI from the map interaction. There was not enough time to do a proof-of-concept implementation, but Roald and Volker are about to do this in a sandbox. Expect more discussion on this on the dev list in the near future.
Bart van den Eijnden (OSGIS) and Eric and reviewed some tickets that were in the pipeline and put the fixes into trunk.
Having developers of several projects sitting in the same room also leads to synergies: Klokan Petr Přidal (known from MapTiler), as he was sitting next to Mike Adair (known from proj4js) coded a quick example of raster reprojection with Canvas and proj4js. Klokan Petr Přidal showing Mike Adair and Marc Jansen that proj4js can do more than just vector reprojection.
I (Andreas Hocevar, OpenGeo) did my best to support everyone who participated. Apologies if I wasn’t able to help with all issues during the sprint, but I will be available with support and advice for all the good work that was started, so it can be finished successfully.
We’ve seen a good bit of OpenLayers activity at the conference so far. Last night, a crowd of 25 or so gathered at the OpenLayers Birds of a Feather session. It was a good opportunity to hear how people were using the library in their applications. We got plenty of suggestions on ways to improve things. Among them:
More OpenLayers.Console.warn and error calls for better diagnostics in debug mode.
Clean up examples: more consistent coding style, remove examples of deprecated behavior (or archive them separately), organize examples by category and make them more easily discoverable.
We’ve been advancing the editing capabilities of OpenLayers with each release since the initial inclusion of the vector feature rendering in 2.4. The upcoming 2.8 release will bring two new enhancements that I’m particularly excited about: snapping and splitting.
The snapping example demonstrates some of the configuration options of the snapping agent. This control can be set up to work with any number of target layers. Modified features in a source layer are snapped to eligible features in target layers based on tolerance settings. Snapping options for nodes, vertices, and edges can be independently configured.
The split feature example uses the new split control to generate temporary sketches for splitting existing features. The control can also be configured to listen for edits on an existing layer and split features on any target layer.
The snapping and split controls can be used together to provide an editing environment that maintains some basic topology rules. Both controls trigger events that let the application designer customize exactly what happens before, during, and after snapping and splitting. You can find documentation and additional examples for both of the controls in the growing prose documentation.
Thanks to Sweco for supporting this work. We’re excited to have it in the trunk.
Here’s an OpenLayers Holiday Present, for those of you working on getting OpenLayers in more places in your organization:
OpenLayers Case Studies: Examples showing the usage of OpenLayers inside of organizations, and how the use of OpenLayers has helped that organization.
Currently, the number of case studies is small, but we’re working on growing these case studies so that for any particular application, there is an example of someone who has used OpenLayers to do something similar.
If you’re interested in sharing a case study with the OpenLayers project, feel free to comment here, or follow up to the mailing list post made on the topic.
In the November issue of the GeoConnexion magazine, (”Geo: International”), an article I wrote was published in OSGeo’s Monthly Column, “Open Sources”. The article talks a bit about the history of OpenLayers, and how it came to be developed the way it did:
OpenLayers was the first mapping framework to make an explicit statement that it was not an application at all, but a toolkit for building mapping applications. This different approach resulted in a somewhat long curve to acceptance. In its infancy, the project was used only by developers: people who had a strong knowledge of what they wanted to do, and needed to have more control over their tools in order to do it. This early audience helped to build a rapid development environment where many of the users of the code were also able to contribute fixes and improvements based on their needs. This developer-friendly environment may be one of the key differences that has allowed OpenLayers development to continue to grow.
We had a good time in Cape Town putting on an OpenLayers intro workshop. The 3 hour session was well attended, and very well supported by OpenLayers developers. Thanks to those who participated and especially to all the developers that came to help out.
I have made the workshop documents available - and am looking forward to putting up a couple translations soon (French and Spanish at least). I’ll be shifting around the structure a bit and making an HTML version (and RST source) available at some point. If you’re interested in using the material, please do - we’ve put on a Creative Commons Attribution-Share Alike license.
You can find some instructions on getting set up for the workshop on our workshop wiki. Note this version requires a local GeoServer running (or a proxy). I’ll update the material to work with a publicly available GeoServer at some point.
If you make use of the material, or are interested in doing a translation, let me know (me @ opengeo.org).
The OpenLayers Development Team is proud to announce the first release
candidate of OpenLayers 2.7!!
As of 2.7-RC1, the OpenLayers 2.7 release closes 187 outstanding
tickets, split nearly down the middle between bug fixes and new features. Although this ticket count is less than previous releases,
the 2.7 release is the first release which has been executed as a
date-based release (instead of feature-based)… which represents a
bit of a departure for the project. Whether or not we will continue to
make releases based on this fixed date style is up in the air, but the
experiment has proved interesting and mostly successful.
The 2.7 release should not present the drastic changes that we’ve seen
in previous releases (like 2.6). Overall, it turns out to be much more
of a bug fix release, although it does include some exciting new
functionality. Highlights include:
* Vector-Behavior: Strategies, Protocols, Filters (though the
completion of this going in at RC2)
* Improved Vector rendering for better performance
* Canvas rendering class
* Z-Ordering and Y-Ordering for Vector layers
* New Basic Measurement Control
* New OpenLayers.Request interface for AJAX
* Smarter Popups
For information on possible changes that will need to be made between
this version of OpenLayers and previous versions, please look at the Release notes.
We invite you to help us test the 2.7 release candidate! To test 2.7
in your applications, include the following tag in your
OpenLayers-powered page:
As always, the source is available at http://openlayers.org/download/.
Bug reports can be filed in Trac, under the 2.7 version and 2.7 RC1
milestone.
There are so many people who have really made this release happen, too
many to name here. OpenLayers is not about the fame, it’s about the
glory. You know who you are. Thank You.
A special thanks, though, goes out to John Frank/Josiah Strandberg at MetaCarta and Chris Holmes from OpenGeo for putting up the funds and
clearing away the time so that we could have an excellent ‘bunker
week’ out in Cambridge at the end of July. That bunker was a critical
pow-wow for us in terms of getting things under control for the 2.7
release and also in general for project synergy and getting some great
patches in. Thank you!
All forthcoming RC announcements will be sent only to the Developers
list: anyone interested in tracking the progress to a final release
should subscribe to that list.
Last week, a number of OpenLayers developers gathered in Cambridge, MA to push towards a 2.7 release. Erik Uzureau and Chris Schmidt from MetaCarta hosted. Andreas Hocevar, Sebastian Benthall, Tim Coulter, and I went from the OpenGeo team. Closing tickets against 2.7 was our main objective, and we managed to get a handful of cool features in.
Tim C. and Erik worked together to integrate Tim’s nice y-ordering support for vector features (see the example with background shadows). Andreas continued to make vector rendering and styling improvements with the addition of well-known graphic names and dash style. Chris closed a good number of tickets and made some performance enhancements. Seb and I laid the groundwork for bringing in the new vector layer behavior work (strategies and protocols).
Thanks to Erik for all the organizational work, to MetaCarta for hosting us in their office and providing time for Erik & Chris, and to The Open Planning Project for supporting the rest of us in the sprint.
