- From: Matt Giuca <notifications@github.com>
- Date: Tue, 30 Jun 2020 22:34:34 -0700
- To: w3c/manifest <manifest@noreply.github.com>
- Cc: Subscribed <subscribed@noreply.github.com>
- Message-ID: <w3c/manifest/pull/834/c652201560@github.com>
Doing a last-minute re-read. There's a bit of a complication (as usual, because of the unfixable awkwardness that `scope` is based on `start_url` and not the other way around): 1. This PR introduces the possibility that `start_url` of a processed manifest is undefined. (Previously, this was impossible; a missing or invalid `start_url` would default to the document URL. Note that fixing this is the whole point: since there are cases where you don't _have_ a document URL, this was unimplementable.) 2. The "processing the `scope` member" is unchanged by this PR. It assumes _start URL_ is a valid URL. 3. If _start URL_ is missing from these steps, Step 1: _default_ will be **failure** (it is illegal to URL-parse "." without a base). 4. Step 5: "within scope" isn't defined when the URL is undefined. We'll assume it returns false in this case. 5. Any attempt to return _default_ will mean that `scope` of a processed manifest is **failure**. So if `start_url` is omitted, the manifest processing kinda explodes. The simplest fix for this would be to simply say "if _start URL_ is undefined, `scope` is undefined". That raises a follow-on possibility of what things will do when `scope` is undefined. (For example, "within scope of a manifest" is meaningless when `scope` is undefined due to `start_url` being undefined, though that one's easy to fix by simply making it always false.) Need to think a bit more about this before landing. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/w3c/manifest/pull/834#issuecomment-652201560
Received on Wednesday, 1 July 2020 05:34:47 UTC