You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Siblings of #4941: WEB SEND server options still refused (CHUNKING, CONTAINER/CHANNEL, code pages, CLOSESTATUS) #5037
Siblings of #4941 (class: WEB server-side commands, #4767). Found in the review of PR #5036, which models WEB SEND ACTION(EVENTUAL) / DOCSTATUS and WEB RETRIEVE NORMAL.
These WEB SEND (server) options are still refused by name in gitgalaxy/standards/cics/commands/web.py (register X37). IBM documents each of them on the WEB SEND (server) page (https://www.ibm.com/docs/en/cics-ts/6.x?topic=summary-web-send-server). No burned-estate program uses them today, so this is low priority.
WEB SEND CHUNKING(CHUNKYES / CHUNKEND): chunked transfer-coding. IBM's default ACTION for a chunked send is IMMEDIATE.
WEB SEND CONTAINER / CHANNEL: a response body taken from a channel container.
WEB SEND CHARACTERSET / CLNTCODEPAGE / HOSTCODEPAGE: code-page conversion of the response.
WEB SEND CLOSESTATUS(CLOSE / NOCLOSE): closing the connection after the response.
Review of PR #5124 (tick 2026-10-14): CONTAINER / CHANNEL modelled; CHUNKING, CHARACTERSET / CLNTCODEPAGE / HOSTCODEPAGE and CLOSESTATUS keep their refusal with the IBM-sourced reason now named in X37 (this issue's "or keep the refusal and name the reason" path). New member found in the sweep, added here rather than filed: a DATATYPE BIT container sent without conversion (refused: needs a non-text body comparison). DFHTRANSACTION as CHANNEL stays under X17; client-side WEB SEND is X35 (different mechanism).
PR #5124 models WEB SEND CONTAINER/CHANNEL per IBM and keeps CHUNKING, CHARACTERSET/CLNTCODEPAGE/HOSTCODEPAGE and CLOSESTATUS refused by name with IBM-sourced reasons in register X37. The issue allowed either route, so the acceptance is met.
Activity
CICS: WEB SEND CONTAINER / CHANNEL as the response body; CHUNKING and…
squid-protocol commentedon Oct 14, 2026
Review of PR #5124 (tick 2026-10-14): CONTAINER / CHANNEL modelled; CHUNKING, CHARACTERSET / CLNTCODEPAGE / HOSTCODEPAGE and CLOSESTATUS keep their refusal with the IBM-sourced reason now named in X37 (this issue's "or keep the refusal and name the reason" path). New member found in the sweep, added here rather than filed: a DATATYPE BIT container sent without conversion (refused: needs a non-text body comparison). DFHTRANSACTION as CHANNEL stays under X17; client-side WEB SEND is X35 (different mechanism).
CICS: WEB SEND CONTAINER / CHANNEL; named IBM refusals for the rest (…
squid-protocol commentedon Oct 14, 2026
PR #5124 models WEB SEND CONTAINER/CHANNEL per IBM and keeps CHUNKING, CHARACTERSET/CLNTCODEPAGE/HOSTCODEPAGE and CLOSESTATUS refused by name with IBM-sourced reasons in register X37. The issue allowed either route, so the acceptance is met.