syndicate-lang
/
syndicate-rkt
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Remove unused definition

This commit is contained in:
Tony Garnock-Jones 2019-05-07 14:02:52 +01:00
parent 39906803b2
commit 88ec80487d
1 changed files with 25 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

27
syndicate/distributed/client/tcp.rkt
View File

@ -19,8 +19,6 @@
(during/spawn (server-connection ($ address (server-tcp-connection $host $port $scope)))
#:name address
(define id (list (gensym 'client) host port))
(let boot-connection ()
(define root-facet (current-facet))
(reassert-on (tcp-connection id (tcp-address host port))
(retracted (tcp-accepted id))
@ -33,20 +31,19 @@
(on (message (tcp-in id $bs)) (accumulate! bs)))
(during (server-transport-connected address)
;; If we run generic-client-session-facet in the `tcp-accepted` handler above,
;; then unfortunately disconnection of the TCP socket on error overtakes the error
;; report itself, terminating the generic-client-session-facet before it has a
;; chance to handle the error report.
;; If we run generic-client-session-facet in the `tcp-accepted` handler above, then
;; unfortunately disconnection of the TCP socket on error overtakes the error report
;; itself, terminating the generic-client-session-facet before it has a chance to
;; handle the error report.
;;
;; Could timing errors like that be something a type system could help us with?
;; The conversation in `server-packet`s is sort-of "nested" inside the
;; conversation in `tcp-in`s; a single facet reacting to both conversations (in
;; this instance, to `server-packets` in an implicit frame, but explicitly to the
;; frame of the `tcp-in`s, namely `tcp-accepted`) is probably an error. Or rather,
;; any situation where pending "inner conversation" business could be obliterated
;; by discarding a facet based on "outer conversation" framing is probably an
;; error.
;; Could timing errors like that be something a type system could help us with? The
;; conversation in `server-packet`s is sort-of "nested" inside the conversation in
;; `tcp-in`s; a single facet reacting to both conversations (in this instance, to
;; `server-packets` in an implicit frame, but explicitly to the frame of the
;; `tcp-in`s, namely `tcp-accepted`) is probably an error. Or rather, any situation
;; where pending "inner conversation" business could be obliterated by discarding a
;; facet based on "outer conversation" framing is probably an error.
;;
(generic-client-session-facet address
scope
(lambda (x) (send! (tcp-out id (encode x)))))))))
(lambda (x) (send! (tcp-out id (encode x))))))))