If done: lnk_db_conn() and fresh::frs_db_conn() connect to the same database in the same environment. If never: on hosts that set both groups, a bare lnk_db_conn() goes to the PG_*_SHARE target (the bcfishpass tunnel) while frs_db_conn() goes to the PG* target (local fwapg). A pipeline mixing the two helpers reads from one DB and writes to another.
fresh#213 (fresh 0.36.0) changed frs_db_conn() to resolve explicit args → standard libpq vars → legacy PG_*_SHARE (deprecated, message once) → libpq defaults, and never mixes the groups. lnk_db_conn() (R/lnk_db_conn.R:38) still does PG_*_SHARE first, then PG*, then hardcoded defaults, per field. Its roxygen says it "works identically to frs_db_conn()", which is no longer true.
Proposed: have lnk_db_conn() delegate to fresh::frs_db_conn(), or mirror its resolver. Bump the fresh minimum to 0.36.0.
If done:
lnk_db_conn()andfresh::frs_db_conn()connect to the same database in the same environment. If never: on hosts that set both groups, a barelnk_db_conn()goes to thePG_*_SHAREtarget (the bcfishpass tunnel) whilefrs_db_conn()goes to thePG*target (local fwapg). A pipeline mixing the two helpers reads from one DB and writes to another.fresh#213 (fresh 0.36.0) changed
frs_db_conn()to resolve explicit args → standard libpq vars → legacyPG_*_SHARE(deprecated, message once) → libpq defaults, and never mixes the groups.lnk_db_conn()(R/lnk_db_conn.R:38) still doesPG_*_SHAREfirst, thenPG*, then hardcoded defaults, per field. Its roxygen says it "works identically tofrs_db_conn()", which is no longer true.Proposed: have
lnk_db_conn()delegate tofresh::frs_db_conn(), or mirror its resolver. Bump the fresh minimum to 0.36.0.