Closed robyww closed 2 years ago
One more thing - this issue also exist in a simple data link url as well.
Hum, not sure I catch your point actually. If the content_type is "text/html" (and I confirm that the new version will have and optional "content-type" PARAM in the descriptor) what other choice can we have to send the link to the browser ? What you ask is to describe this use case explicitly in the content_type subsection ?
I think I am telling you what we had to do. My issue is there is not a way besides some sort of content-type guess to tell the difference between a data product and a web page that should be opened. If content-type is the right way to do it then it makes sense to call it out. However, I could not tell what the right way to do it is. This seems to me to be a important way to use service descriptors that might have been overlooked in the past.
I think I am telling you what we had to do. My issue is there is not a way besides some sort of content-type guess to tell the difference between a data product and a web page that should be opened. If content-type is the right way to do it then it makes sense to call it out. However, I could not tell what the right way to do it is. This seems to me to be a important way to use service descriptors that might have been overlooked in the past.
Roby, close to publishing the 1.1 WD see how we can solve this one. In case content-type tells its html we can say the behavior is always to push that to the browser. Then the user will discover if it is downloadable html page (let's say a preview for example) or any kind of interactive interface. Should that solve that issue for you ?
I think so.
I think so.
Attempt to solve in PR #69
clarified in doc
Most service descriptors specify data products that can be downloaded and possibly visualized. However a powerful application of service descriptors is to specify web pages or web apps. In this case the consumer for the service descriptors needs a way to know that the url is not to be downloaded but should be opened in the web browser.
Currently we are looking at the
content_type
to see if it istext/html
. If this is the best solution then it would be good if the standard could specifically call it out.