parser-state: fix attribute merging #11294
Conversation
github-actions
bot
added
the
with-tests
| @@ -189,6 +155,60 @@ inline void ParserState::addAttr(ExprAttrs * attrs, AttrPath && attrPath, const | |||
| } | |||
There was a problem hiding this comment.
I don't know if this stuff should be moved to the new overload or if it should be left here; I don't really understand what it's doing.
There was a problem hiding this comment.
What stuff are you referring to?
There was a problem hiding this comment.
Sorry, I was referring to the above five lines involving lexerState.positionToDocComment; it's the same ‘stuff’ I'm wondering about in #11294 (comment).
| ae->attrs.clear(); | ||
| jAttrs->dynamicAttrs.insert(jAttrs->dynamicAttrs.end(), | ||
| std::make_move_iterator(ae->dynamicAttrs.begin()), | ||
| std::make_move_iterator(ae->dynamicAttrs.end())); | ||
| ae->dynamicAttrs.clear(); | ||
| if (ae->inheritFromExprs) { | ||
| jAttrs->inheritFromExprs->insert(jAttrs->inheritFromExprs->end(), | ||
| std::make_move_iterator(ae->inheritFromExprs->begin()), | ||
| std::make_move_iterator(ae->inheritFromExprs->end())); | ||
| ae->inheritFromExprs = nullptr; | ||
| } |
There was a problem hiding this comment.
At the end of this block, everything about ae worth retaining should have been merged into jAttrs. In principle I would think that means ae can be deleted. But because of the doc comment stuff lingering in the other overload, I don't think it's safe to do that. I settled for just clearing vectors instead. (I'm guessing at the right non-leaky stuff to do here, in general; are these move_iterators used correctly?)
I'm confused, though, about what good that doc comment stuff can do if it's being attached to an expression that essentially gets dropped on the floor.
rhendric
force-pushed
the
rhendric/fix-11268
branch
from
c780262
to
66ae0e2
Compare
tomberek
added
the
idea approved
rhendric
force-pushed
the
rhendric/fix-11268
branch
from
66ae0e2
to
85b7453
Compare
|
This pull request has been mentioned on NixOS Discourse. There might be relevant details there: https://discourse.nixos.org/t/2024-08-19-nix-team-meeting-minutes-170/50942/1 |
|
Eelco has the week off. He'll review this later. |
|
@edolstra is back afaik. |
rhendric
force-pushed
the
rhendric/fix-11268
branch
from
85b7453
to
05515c5
Compare
|
@rhendric Can you resolve the merge conflict? Thanks! |
Mic92
force-pushed
the
rhendric/fix-11268
branch
from
05515c5
to
8034589
Compare
|
I rebased. There was no actual merge conflict. |
Motivation
Fixes #11268.
Context
This should be non-breaking. As a side effect, it makes some duplicate attribute error messages nicer (later occurrences point back to previous occurrences rather than the other way around).
I'm working at the edge of my proficiency with C++ here, so please take extra care to verify that how I'm handling pointers and memory management is sensible, because I'm not confident that I'm doing it right.
Priorities and Process
Add 👍 to pull requests you find important.
The Nix maintainer team uses a GitHub project board to schedule and track reviews.