The issue was that we were expecting Vega to always include a "text" key when requesting the text width of a string. For some reason, with the spec from #72 the text key is missing. This PR updates out logic to treat a missing text field as an empty string and return width zero. With this change the spec renders properly:
I also made an update to the test logic to make it easier to add new image baselines in the future. Now when a baseline doesn't exist the computed image is written to the "failed" directory.
Closes https://github.com/vega/vl-convert/issues/72.
The issue was that we were expecting Vega to always include a "text" key when requesting the text width of a string. For some reason, with the spec from #72 the text key is missing. This PR updates out logic to treat a missing text field as an empty string and return width zero. With this change the spec renders properly:
I also made an update to the test logic to make it easier to add new image baselines in the future. Now when a baseline doesn't exist the computed image is written to the "failed" directory.