Ömer Sinan Ağacan

> I wrote a small tool to generate the character set (about 15 kB of text) and include it Is this text full of unique characters, or do you extract...

Given a header it returns event parsers (`mkEventParsers :: Header -> EventParsers`). See the definition here: https://github.com/osa1/ghc-events/blob/5090f824fa709415336248342674fdd4862695da/src/GHC/RTS/Events/Incremental.hs#L161 basically just the event parser builder from `mkEventDecoder` factored out for reuse.

I added a bunch of debug prints and it seems like the parser for EVENT_GC_STATS_GHC event can actually parse this and the parsed evSpec is identical to the original one,...

OK so this is because of `padParser` which for some reason makes this parser expect one more word after the actual contents. Investigating more ...

So according to the event type parser event type of EVENT_GC_STATS_GHC is ``` EventType {num = 53, desc = "GC statistics", size = Just 58} ``` this says the size...

Right, so GHC generates one more word for `EVENT_GC_STATS_GHC` which is for "par_balanced_copied": https://github.com/ghc/ghc/blob/65eec9cfd4410c0e30b0ed06116c15f8ce3de49d/rts/eventlog/EventLog.c#L413-L424 Not sure how to deal with this ...

Quoting from the diff that added this new field: https://phabricator.haskell.org/D3658 > Add the new value at the end, and we'll gradually deprecate par_max_copied. Old clients that expect the old format...

I haven't debugged this myself yet, but my understanding is that this problem occurs because there are multiple ways to parse an event (in other words we map different events...

Here's another use case. Rust integers can be lexed like this: ```rust rule Init { ... ("0b" | "0o" | "0x")? ($digit | '_')* $id? =? |lexer| { let match_...

I'm not sure how to implement this feature yet, but here's an idea. Suppose I have a rule with these regexes: ``` => ..., => ..., ``` In the generated...