00:20 <discocaml> <Et7f3 (@me on reply)> What kind of side effect ? memory allocation (all) ? printing to file ? (function with print in the name theorically and flush if print is called before) opening file, ...

00:21 <discocaml> <Et7f3 (@me on reply)> What kind of side effect ? memory allocation (almost all), printing to file (function with print in the name theorically and flush if print is called before) opening file, ...

00:26 <discocaml> <ilo Kali> memory allocation isn't relevant for what I'd like to do, but yeah; printing, opening/outputting to/inputting from/closing channels, etc.

00:32 <discocaml> <psychokitty> roughest of rough heuristic would be where `unit` is the (co)domain

00:33 <discocaml> <psychokitty> but of course, probably others that do some stuff and provide a value

00:34 <discocaml> <ilo Kali> yeah, that was my thought too

12:31 <discocaml> <Et7f3 (@me on reply)> Unix module is full of side effect but other should be pure

12:31 <discocaml> <Et7f3 (@me on reply)> Unix module is full of side effect but other shouldn't have much

12:50 <discocaml> <leviroth> Does adding something to a hash table count as a side effect?

13:02 <discocaml> <Bluddy> Generally yes. The exception is that a hashtable created within a function and updated within the scope of that function does not seem like a side effect to the outside world.

13:05 <discocaml> <leviroth> I meant for @ilo Kali ‘s purposes

15:23 <discocaml> <ilo Kali> Yeah, hashtable operations would count

22:01 <discocaml> <ec> @cristianoc not sure if you check Discord; but if so — can I DM? to send a copy of the entire file for https://github.com/rescript-association/reanalyze/issues/187, as a test-case. let me know! (=

22:01 <discocaml> <ec> also, _frickin'_ cool tool.

22:19 <discocaml> <Et7f3 (@me on reply)> Can't you reduce it ?

22:20 <Anarchos> in my dune repo, i have the 3.6.2 informations in the Changes.md. But when i compile.install i still get 2.8.0-2534-g902dd01-dirty

22:25 <discocaml> <Et7f3 (@me on reply)> To reduce you can follow method of https://llvm.org/docs/Bugpoint.html remove function, remove function code, remove variable, arguments/parameters, remove field,...

22:29 <discocaml> <ec> Yeah, I'm familar — the problem here, is that it's some sort of location/parsing error. Removing content that *seems* unrelated changes the occurrence of the bug.

22:30 <discocaml> <ec> some content earlier in the file — haven't had a lot of luck narrowing down _what_ content, although I'm working on that right now — seems to cause a loss of position-accuracy; and then Reanalyze ends up writing content into a dislocated position in the middle of a syntax-word a half-dozen lines past where it's trying to write.

22:31 <discocaml> <ec> At least it's deterministic 😓

22:32 <Anarchos> sorry i crashed