libera/#clasp - IRC Chatlog
Search
15:46:44
drmeister
They have a more detailed description of the force field file but they link to the SMARTS and SMIRKS documentation that we already have for details on those.
15:54:48
drmeister
Reading the SMARTS documentation this morning leads me to conclude that these `@` indicate `in cyclic bond`
18:52:09
drmeister
I just got confirmation that the `*@1-[*]=,:[*][*]=,:[*]@1` means "Inside a 5-member ring"
21:06:38
drmeister
::notify scymtym It's exactly as painful as I thought it would be to try and get hold of these people to ask them how this is implemented. They refer me to the SMARTS links that we have already read and then they tell me to look in the rdkit code. I don't think the answer is in there.
21:09:19
drmeister
::notify scymtym I think the answer is buried in the OpenEye library. I am starting to think that the `@1` is redundant. The numbers are used to indicate rings, the `@` indicates a ring. I think `@1` and `1` are equivalent.
21:14:16
drmeister
::notify scymtym I'm really rusty with esrap - if I want to optionally allow an `@` in `(defrule atom-pattern ...)` how would I do that? Instead of `(and acyclic-atom-pattern (? parser.common-rules:integer-literal/decimal))` to allow an optional `@` before the integer?
23:26:22
drmeister
::notify scymtym Is there a dynamic variable available in defrule that provides the entire string that is being parsed?
1:42:29
drmeister
::notify scymtym This illuminates how ring-closures work - you can prefix them with a bond primitive. This extends to SMARTS so things like `*=1[*][*][*][*]=1` are valid but the `*@1[*][*][*][*]@1` still seems pointless.
3:37:17
Bike
yitzi: any idea what's up with these test failures? the closest i got to changing koga was a line in repos.sexp, but for some reason it caused the llvm json thing to pop up again https://github.com/clasp-developers/clasp/pull/1397