Greg Pfeil

Is this actually a round-trip bug? Shouldn’t deleting a unique type fail if there are dependencies on that type?

You can kind of test this sort of thing now, since #5826. ```` markdown ``` unison :hide OnMyFeatureBranch = () ``` ``` ucm proj/my-feature> add Okay, I'm searching the branch...

Here’s the code at that commit: https://github.com/unisonweb/unison/blob/a74d4e8cb9cd860ebd79e348f8012f57625e2239/unison-syntax/src/Unison/Syntax/Lexer.hs#L229-L241 A first step might just be returning a parser failure rather than `error` (and maybe changing the description).

I don’t think so – this has just hopefully made the failure more informative when it occurs again (since it‘s clearly not impossible). Since this is an failure in the...

@ehamberg Sorry to have caused you trouble. Can you show me how you were setting `auto-dark-themes`? The removal of the optional argument to `load-theme` should have been called out more...

> Humm, so this is also happening on Vanilla Emacs? Maybe we should add this 'fix' on the Readme for all cases. What do you think @sellout ? Yeah, `custom-safe-themes`...

Ok, I’ve thought about this more than I should. First, I was wrong to suggest `auto-dark-assume-themes-are-safe`. The current mechanism for themes is pretty good. When you load (a particular version...

Ah, I made all these changes last night. I should have put up at least a draft PR. I’ll get one up tonight, and people can tell me what they...

Ah, thanks @mepcotterell, that context really helps. > If other elisp code maliciously adjusts the load path so that a different theme with the same name is loaded when auto-dark-mode...

I put up #67, which addresses at least some of the concerns here. Didn’t officially say it “fixes” this issue, as there are a lot of different things going on...