Change Details

Currently, spawning a detached process, it assumes no input/output from/to its parent process. To support synchronous invocation of gpg-agent for GnuPG, when input from detached process can be available, parent process could check if gpg-agent is ready for serving with the socket. (1) GnuPG spawns gpg-agent process with stdin+stderr (2) gpg-agent can emit errors to stderr, emit "OK\n" when it's ready, or "ERR <ERRCODE>\n" to its stdout (3) GnuPG checks stdin to see gpg-agent is ready

Currently, spawning a detached process, it assumes no input/output from/to its parent process. To support synchronous invocation of gpg-agent for GnuPG, when input from detached process can be available, parent process could check if gpg-agent is ready for serving with the socket. (1) GnuPG spawns gpg-agent process with pipe input+stderr (2) gpg-agent can emit errors to stderr, emit "OK\n" when it's ready, or "ERR <ERRCODE>\n" to its stdout (3) GnuPG checks pipe input to see gpg-agent is ready (4) gpg-agent closes its stdout and stderr (5) GnuPG closes pipe input For now, I don't know if the scenario above works, let me try.

Currently, spawning a detached process, it assumes no input/output from/to its parent process. To support synchronous invocation of gpg-agent for GnuPG, when input from detached process can be available, parent process could check if gpg-agent is ready for serving with the socket. (1) GnuPG spawns gpg-agent process with stdinpipe input+stderr (2) gpg-agent can emit errors to stderr, emit "OK\n" when it's ready, or "ERR <ERRCODE>\n" to its stdout (3) GnuPG checks stdinpipe input to see gpg-agent is ready (4) gpg-agent closes its stdout and stderr (5) GnuPG closes pipe input For now, I don't know if the scenario above works, let me try.