29 Dec 2023 |
boltless | parser action in state 26931 should be accept instead of resume | 11:39:52 |
| tomcek8 joined the room. | 12:55:09 |
boltless | .vhyrro can we remove free-form inline_macro? as parser generated with nightly tree-sitter works fine with up to two kind of verbatim free-form attached modifiers, if we remove free-form inline_macro, free-form is near complete | 13:12:23 |
boltless | as this is not a runtime bug, we don’t need nightly tree-sitter to use the v3 parser. We can build&run parser in 0.20.8 version without errors after generating parser with nightly version. | 13:16:15 |
boltless | theoretically this would be fixable by editing grammar.json directly | 13:17:12 |
boltless | * theoretically having three different verbatim free-form attached modifier is possible by editing grammar.json directly | 13:18:07 |
| veganiks joined the room. | 13:40:59 |
boltless | solved | 14:24:48 |
.vhyrro | well not really, everything has to have a free-form variant | 14:27:55 |
.vhyrro | not much of one, but it can't be an arbitrary syntax edge case :/ | 14:29:40 |
boltless | I’ll go with current version anyway because tbh I think no one would even notice free-form inline_macro disappeared plus, this is theoretically fixable if there is any brave person who would manually edit grammar.json file | 14:31:54 |
.vhyrro | yep sounds good | 14:32:42 |
.vhyrro | maybe we should generate the whole grammar.json file from scratch for maximum performance | 14:32:57 |
boltless | known issues left with paragraph parsing: - parser size is quite big (3.1MiB without linkables/link modifiers/attached modifier extensions) - missing one of verbatim free-form attached modifier (bug on tree-sitter) - link-modifier breaks the whole parser like verbatim free-form attached modifier does : I have some idea | 14:35:24 |
boltless | that’s good idea. good idea that no one would want to try | 14:36:11 |
| Sir Polo Modulus III changed their profile picture. | 15:18:46 |
boltless | link modifier is done by parsing :* as single (bold_open) token some pros of loosing actual (link_modifier) node from AST - doesn’t affect parser size - can still capture : as link_modifier with query - logic is really simple as it is using pre-existing non_open rule | 17:22:28 |
boltless | * link modifier is done by parsing :* as single (bold_open) token some pros of loosing actual (link_modifier) node from AST - doesn’t affect parser size - can still capture : as link_modifier with query - logic is really simple as it is using pre-existing non_open rule | 17:22:38 |
.vhyrro | also sounds good tbh | 17:22:40 |
.vhyrro | if it can be captured with queries then that's all that matters | 17:22:44 |
boltless | Download IMG_0096.png | 17:40:19 |
boltless | then I can finally confirm attached modifier is complete only thing missing is a free-form inline_macro | 17:40:20 |
.vhyrro | that's amazing | 17:49:49 |
.vhyrro | well done mate | 17:49:54 |
.vhyrro | did you simplify the scanner by any chance? that should be the next TODO if possibel | 17:50:19 |
boltless | so anchor seems to be a next problem | 17:55:03 |
boltless | just solved both anchor and free-form inline_macro wth | 18:26:14 |
.vhyrro | what da hellll | 18:26:33 |
boltless | .vhyrro can you guess the reason? it was _character token for this whole time | 18:26:46 |
.vhyrro | you're taking the piss, seriously? | 18:26:54 |