Discussion - week of 07/21/14

[ezoehunt]

This is where we'll talk this week.

[dbialer]

Interesting tests results. Though hard to make a direct correlation from these user tests with data from submission website, seems to validate that a larger percent of app submission validations fail validation. Creating manifests is difficult for many developers, especially with packaged apps that then need to be edited, repackaged, and retried. If there is a way to auto-generate some of the manifest, that might help.

[brampitoyo]

Many respondents have remarked that starting from a sample file (versus from scratch) will net them more correct end results. We specifically chose not to test a sample file here, because we want our respondents to read the instructions and see whether we can make improvements to the existing MDN article.

Once they get a helpful sample file to start with, our validation and correctness rate might jump up. This might be an interesting research to do later, when we want to validate our chosen approach.

[eviljeff]

"Less than 10% named the file correctly as manifest.webapp. This is likely to be a major source of manifest validation failure."
for hosted apps the filename isn't a requirement - its the mimetype which is the issue. Although some webservers may already be set up to serve .webapp with the correct "application/x-web-app-manifest+json", its not guaranteed, and its perfectly possible (and acceptable) to serve any random filename with the correct mimetype if the server is setup to do it.

[ezoehunt]

Thanks for the information. That information can be added to the instructions on MDN.