mdavidsaver

> * `pyepics (epicscorelibs), aioca, p4p`: https://github.com/coretl/can_epics_clients_coexist/actions/runs/4616749420/jobs/8162161081 This is a crash on exit via `Py_FinalizeEx()` and `atexit_callfuncs()` ending in an libffi (ctypes) call to `ca_clear_channel()`. So either pyepics or aioca....

> ... Do you know how to destroy a different thread's ca_context? This can't be done as such. Instead, imo. a library/module using libca contexts should take steps to create...

> > I've automated this as https://github.com/mdavidsaver/ci-core-dumper > > I added *[ci-core-dumper]*, and it works well for Linux, but doesn't appear to do anything for Windows or Macos, am I...

> It looks like epicscorelibs does not package libraries for pvDatabase, pvaClient, normative types, that are all required by pvapy. Prior to `epicscorelibs==7.0.3.99.4.0` the pva2pva module was not included. Likewise...

> ... Michael's implementation of the QSRV server ignores that and sends everything ... To reiterate... The pvRequest mask is not ignored. Rather, it is interpreted differently by QSRV. The...

Maybe because `lev` is outside the loop? Also, as I read it, `lev` is never set back to 0. So because this variable is outside the loop, won't this apply...

Initializing to `1` is a improvement. I think it should be done inside the loop though. ```c while(fgets(inbuf,sizeof(inbuf),fd)) { ++line; pattern=rname=hname=NULL; lev = 1; /*

> I don't see anything in the CA protocol description which _requires_ a client to send a `CA_PROTO_SEARCH` before sending a `CA_PROTO_CREATE_CHAN` message with the same name ... wrt. PVA....

> You are right about the chances of playing ping-pong with an IOC by asking for a PV with a wrong field name or filter. To my mind, a better...

> ... that would create many millions of legit names. Right, I was thinking (but not writing) to match prefix on record names. That is why I was thinking about...