commit 5e748f4e7c8884ea654425045f9cdc4e17c5cb06 parent 3d06ab587f00e19f6da6fd2c544a3584f267705a Author: MS <ms@taler.net> Date: Mon, 23 Oct 2023 12:38:58 +0200 DD 50 Q&A Diffstat:
| M | design-documents/050-libeufin-nexus.rst | | | 5 | ++++- |
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/design-documents/050-libeufin-nexus.rst b/design-documents/050-libeufin-nexus.rst @@ -311,4 +311,7 @@ Discussion / Q&A initiated flag in the table to true". May there be a case where the server accepted the request, but the client never got any response (some network issue..), and therefore didn't set the submitted flag, ending - up in submitting the payment twice? + up in submitting the payment twice? Also: flagging the payment _after_ + the bank response, may lead double-submission even if the HTTP talk ended + well: it suffices to crash after having received a "200 OK" response but + before setting the submitted flag to the database.