URL parse errors for `std:none|http://:broken` and `http://:broken` are handled differently

drufball / layered-apis

A new standards effort for collaborating on high-level features.

275 stars 12 forks source link

URL parse errors for `std:none|http://:broken` and `http://:broken` are handled differently #18

Closed hiroshige-g closed 6 years ago

hiroshige-g commented 6 years ago

URL parse error inside layered API fetching URL, e.g. std:none|http://:broken, is handled a little differently from usual URL parse errors (e.g. http://:broken).

<script> error events for std:none|http://:broken are (1) fired in #execute-a-script-block.
<script> error events for non-URL-parsing reasons are (1) fired in #execute-a-script-block.
<script> error events for http://:broken are (2) fired asynchronously, immediately after #prepare-a-script.

hiroshige-g commented 6 years ago

I'm not sure whether:

[A] we should fire all <script> error events in (1) (i.e. fix error events for http://:broken as a non-LAPI-related HTML spec issue), or [B] fire all URL-parser-related <script> error events in (2) (i.e. fix error events for std:none|http://:broken inside LAPI spec).

Fixing issue #16 will be helpful for [B].