Where next for qgis2web?
Those of you who follow QGIS closely might have noticed that development of qgis2web has significantly slowed since QGIS 3 was released. Is this of concern? What next for the plugin?
Slow and steady wins the game?
Perhaps all is well. Perhaps the fact that not many bugs are being found simply means that less development is required, and we don’t need to worry. This does not seem likely. qgis2web is simply middleware which bridges amazing software which itself has since developed significantly. A lack of qgis2web development suggests that new functionality and better performance is not being exploited.
Longstanding issues
Some bugs and feature requests have been around for a long time. Notable among these is the almost non-existent support for layer groups. While the majority of people thankfully do not overload their webmaps with layers, a significant minority do need this functionality.
Structural problems
While the GUI code is reasonably elegant, much of the export code is not, in three major ways:
- not object-oriented
- poor string handling (lots of concatenation rather than templating/formatting)
- output JS fragments insufficiently separated out from Python
These three issues make the code clunky, difficult to understand, and far from well conceived or engineered.
Shiny new features
In addition to poor code and missing basic features, however, there are more interesting new areas which could be further developed:
- export to Mapbox GL JS to bring the power of WebGL to your webmaps, allowing users to create fast, smooth vector maps without coding or hosting with Mapbox
- further develop the work by Paul dos Santos which allows the look and feel of the webmap to be customized
- though not qgis2web, take the idea behind web2qgis and develop it into something fully realized
- make use of QGIS 3 task manager to improve performance
Dependency updates
Two major updates are probably now due:
- OpenLayers 5
- migrate GUI from QtWebkit to QtWebEngine
These are fundamental changes, but ones which probably attract less attention. The longer both are left, the more qgis2web will be held back from future development.
Update: migrating to QtWebEngine seems currently premature at best, and at worst impossible.
Second update: upgrade to OpenLayers 5 has started, and is perhaps less of a big job than anticipated.
Documentation, documentation, documentation
“Scant” is a reasonable description of qgis2web’s documentation. Thanks to work by Herb Fargus, what is there is largely correct and up-to-date, but so much more could be documented.
Time and expertise
If you think that you could contribute to any of this, your input would be incredibly appreciated. Go to the qgis2web bugtracker and see what takes your fancy.