Sender | Message | Time |
---|---|---|
2 Jan 2024 | ||
slobby_bobby joined the room. | 04:39:21 | |
boltless | free-03 case gets really interesting
current output: one open_conflict starting from column 0first * is parsed bold_open second * can’t be bold_open as not_close right after whitespace has higher precedence & we actually opened bold beforeas second * is not bold_open , are remaining punctuations become just punctuation without special meaning | 10:09:33 |
boltless | nvm fixed by giving bold_open higher precedence | 10:17:31 |
boltless | conflict conflict bold in *word /*word* may be parser friendly and it is definitely possible.but error on unclosed is way parser friendly in that context. it just feels like a unintuitive edge case for LTR. I would prefer bold containing one or two conflict for this as it makes more sense, and is also possible + as scanner knows the order of opened modifiers, it would be easier to fix edge cases discovered in the future | 11:54:45 |
boltless | hmmm how can I make test case for this:
parser success with single verbatim for this case, but it did successfully parsed version of two open_conflict as well, but just didn’t selected it for select earlier symbol rule | 14:27:05 |
boltless | we need stack for attached modifiers anyways
parser doesn’t know if bold was opened as free-form or not, so it can’t determine if |* would be valid free_form_close or not | 16:49:23 |
boltless | * we need stack for attached modifiers anyways
parser doesn’t know if bold was opened as free-form or not, so it can’t determine if |* would be valid free_form_close or not when parsing _fail_close | 16:50:19 |
rilplus joined the room. | 21:11:20 | |
santiagomunevar joined the room. | 22:05:05 | |
3 Jan 2024 | ||
boltless | note about current spec: parser will prefer earlier open/close modifiers (LTR) even verbatim can fallback to punctuation with this rule (e.g. *`word* is bold with punctuation and word)but when user opened free-form attached modifier, parser will continue parsing so while *|/word|* is bold, */|word* isn’t as there is valid free-form-italic open after bold open | 03:52:19 |
boltless | .vhyrro do you think this is ok? | 03:53:03 |
.vhyrro | Yeah I see no issues with that | 07:25:53 |
boltless |
now I feel non-verbatim free-form attached modifier is just bad design as free-form opener and free-form closer can be conflicted in various cases | 10:27:04 |
boltless | another free-form edge case is:
as we allows bold_open inside verbatim in case it fails, parser thinks this as unclosed verbatim with unclosed free-form bold inside (which is an error) | 10:36:33 |
boltless | * another free-form edge case is:
as we allow bold_open inside verbatim in case it fails, parser thinks this as unclosed verbatim with unclosed free-form bold inside (which is an error) | 10:36:55 |
boltless | .vhyrro how do you think removing non-verbatim free-form attached modifiers 👀 I think this is fixable, but it would be really v1-style | 10:39:42 |
boltless | or disallow any nested markups inside both verbatim/non-verbatim free-form attached modifiers | 10:48:46 |
boltless | this would be useful for usecases like vim modeline | 10:50:31 |
merole joined the room. | 11:10:55 | |
.vhyrro | in this case i'd take the left-first approach. If it looks like an opener, it's an opener, always | 12:18:58 |
.vhyrro | i've been considering ways of fixing the whole idea of free-form modifires | 12:19:29 |
.vhyrro | but so far I haven't managed to figure out anything truly coherent | 12:19:35 |
.vhyrro | if mrossinek ever comes back to life perhaps we could discuss ;) | 12:19:48 |
boltless | another thing really bothers me building precedence rules is linkables. How can we prevent injections made by linkables? I’ve written some examples in edge case thread in norg-spec repo. | 12:22:04 |
boltless | I think Norg is left-close-first so first closing modifier precedes than any opening modifiers this also explains how unclosed attached modifiers work; parent markup’s closing modifier comes earlier than unclosed one. also with that, user can never make “ /“ bold | 12:51:13 |
4 Jan 2024 | ||
ntbbloodbath set a profile picture. | 02:42:26 | |
ntbbloodbath changed their profile picture. | 06:30:34 | |
avrsarma joined the room. | 07:11:29 | |
.cayuse changed their display name from Cayuse to .cayuse. | 16:48:47 | |
5 Jan 2024 | ||
ntbbloodbath changed their profile picture. | 04:04:40 |