Conversation

Fixing a data race by locally copying const data to allow side-effects (on copy), reminded of this C++ talk:

Replying to

I don't follow the logic in that talk; seems like funcs called from lib forbidden to *modify*, not forbidden to thread-unsafe-modify.

Replying to

I think Herb is arguing that means *logically modify* (i.e. non-const ops), i.e. stdlib objects guarantee const implies threadsafe

Replying to

I thought the section it reffed (which doesn't define modify but uses it) was talking about binary modify, but I'll have to recheck.