2026/04/02

Thursday, April 2, 2026 (#92)

Wednesday Thursday Friday posts: prev < this > next

References Ferreteria/v0.6/clade/IO/Aspect/Connx/run/Starter/Remote/SSH2/@removed ‎ (← links) Ferreteria/v0.6/clade/IO/Aspect/Connx/run/Starter/Local/Proc/@removed ‎ (← links) Ferreteria/v0.6/sys/Connx/run ‎ (← links) 2026/04/01 ‎ (← links) 2026/04/03 ‎ (← links)

• 09:00~ : local schema listing re-achieved, woot! Now to tidy up and then also try to get it working over ssh2 again...
• 10:39 Time to refer back to the Process Control Model to figure out how ssh2 is supposed to operate.
• 10:45 Items to keep in mind:
  • In an SSH2 session (unlike a Proc session), there are two non-Stream resources to keep track of: the ssh2_connect() resource and the ssh2_exec() resource.
  • The Starter object keeps the ssh2_connect() resource, but the Session object needs to pass that to ssh2_exec().
  • Q1: How does the Session object access the Starter object?
• 12:55 A1: Maybe it should work like this:
  • The Starter doesn't do anything except create a Session.
  • The Session opens and closes the ssh2 session, and tracks both that (ssh2_connect()) resource and the ssh2_exec() resource.
    • Q2: Does the ssh2_exec() resource need to be closed as well? (If so, then maybe this ^ isn't a good model, because we'll be opening and closing two separate things.
      • A2: Tentatively, this does not appear to be the case. The documentation does not mention any need to close the resource, nor does there appear to be a function with which to do so.
• 20:34 I think I need a new Process Control Model table, to show what each of the components (Starter, Session) are supposed to do.

Model Action	Action: Proc Starter	Action: Proc Session	Action: SSH2 Starter	Action: SSH2 Session
Open	NO OP		create Session object	ssh2_connect()
Command	proc_open()	send to stdin	NO OP	ssh2_exec()
Shut	NO OP		zap Session object	ssh2_​disconnect()

• 21:55 Trying to document stuff has been almost impossible for the last hour or so, because of cloud5 being overloaded to the point where pages refuse to load many times in a row, requiring 10-20 minutes per edit.
• 21:59 ...and then suddenly everything is fine...