Greatly improve error messages on nightly#525
Conversation
|
I invited you to an experiment I'm currently working on. The want to refactor the parser a lot -- that is, after the current version working version is released. This will change the way how spans are collected quite severely, too. |
|
Well, the parser is the least impacting part here. What I want to change the most is the generator. 😆 |
|
So about the current situation: do you want to postpone this work for after next release and I wait for you to be done on the new lexer before updating the generator or can I finish the current changes and then we'll add the new lexer? |
|
The PR is an improvement over the current state! The new lexer might replicate/replace your work to take the span, but given that it will surely take a while until my changes land, a good thing is a good thing even if it only lasts for a few months / is improved upon afterwards. :) So, no need to wait for me. (I did not look at the PR properly when I posted my previous comment.) |
|
Perfect, then whenever I have time, I'll finish the current PR and send a follow-up for the generator (and the generated code). |
1b9dd0b to
825390d
Compare
|
PR is ready for review. Although I have no clue what's wrong with the CI Book task. Any idea @Kijewski? |
825390d to
52ea474
Compare
52ea474 to
39e2ded
Compare
|
Restarting the CI fixed it. Oh well, dark magic. |
|
PR is ready and CI passed. :) |
Part of #420.
This PR (not ready yet, need to check the
in-docfeature display regression) emits better errors whenSpaninformation is available. Currently it gives:With this PR it gives:
Considering how big it is already, I'm not gonna add span information into the generated code in here, gonna be a (likely massive again T_T) follow-up.