Anton Korobeynikov
Anton Korobeynikov
Tagging @fruffy
> Could you please give a use-case for that? Just loud thinking – if one would need to have internalized string keys (e.g. for faster lookup), but definitely would worth...
I collected some statistics on downstream project. And the results are overwhelming. Essentially cstring cache works as sink, collecting everyone and their brother: all format strings, pieces of input, json...
> One interesting thing could be to (temporarily?) disable[1](#user-content-fn-1-76dde2f097c5f79ec7c36b0899ee2123) the `operator+`/`operator+=` for `cstring`... this is almost never a good idea to use as there is a high chance that an...
> Yes, they are (currently) not garbage collected, but the total size of all cstrings is trivial compared to the total heap size -- generally less than 0.01% I'd be...
Tagging @jafingerhut and @ChrisDodd
Another part missed here from downstream: type inference check that types of loop index variable is compatible with the range.
> I see. I've noticed the Z3 problem with libGC but did not investigate it further. It is good that it can be fixed on libGC side. As far as...
> I also had that issue with linking P4C against other frameworks, for example gRPC. Unclear what causes it but bumping the GC version does seem to fix it. On...
Can it use intrusive reference counters? Instead of shared_ptr wrappers?