)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"f5ba67164d7d8fb74486f87e8947734c75a3d260","unresolved":true,"context_lines":[{"line_number":6,"context_line":""},{"line_number":7,"context_line":"applier: add linear_flow option to avoid O(n^2) graph overhead"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"Taskflow\u0027s graph_flow engine performs a full graph re-analysis"},{"line_number":10,"context_line":"after each action completion, resulting in O(n^2+) complexity"},{"line_number":11,"context_line":"for large action plans."},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"This patch adds a new \u0027flow_type\u0027 config option to"},{"line_number":14,"context_line":"DefaultWorkFlowEngine. When set to \u0027linear\u0027, actions are"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"59557626_a2150e1a","line":11,"range":{"start_line":9,"start_character":1,"end_line":11,"end_character":23},"updated":"2026-03-24 13:32:45.000000000","message":"this was defintly true before \n\nhttps://github.com/openstack/taskflow/commit/6fc3a9b0f1868e3a6092007bb0fd8e79df6f5d95\n\nbut with the visited_nodes caching we shoudl not visit each node only once\n\ndid you test with that fix in place?\n\nwe made that chagne in taskflow based on watcher performacne slaing test we did at the end of q4 last year but the first release of watcher with this will be the 2026.1 release in a few weeks. the first taskflow release with this optimisation is 6.1.1 which was released on febuary 6th of this year.","commit_id":"a8d4fa29ddf0b9fd4c520bbe837f3bc3ff451533"},{"author":{"_account_id":35988,"name":"Hasan Acar","display_name":"Hasan Acar","email":"hasan.acar@tubitak.gov.tr","username":"hasanacar"},"change_message_id":"887b5e884148dae51b2d7ce8c182d1ffef41f823","unresolved":true,"context_lines":[{"line_number":6,"context_line":""},{"line_number":7,"context_line":"applier: add linear_flow option to avoid O(n^2) graph overhead"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"Taskflow\u0027s graph_flow engine performs a full graph re-analysis"},{"line_number":10,"context_line":"after each action completion, resulting in O(n^2+) complexity"},{"line_number":11,"context_line":"for large action plans."},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"This patch adds a new \u0027flow_type\u0027 config option to"},{"line_number":14,"context_line":"DefaultWorkFlowEngine. When set to \u0027linear\u0027, actions are"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"cdb51507_d4364fb7","line":11,"range":{"start_line":9,"start_character":1,"end_line":11,"end_character":23},"in_reply_to":"59557626_a2150e1a","updated":"2026-03-24 14:52:41.000000000","message":"Thank you for the context. I tested with taskflow 6.1.1 and can confirm\nthat the issue is resolved with the visited_nodes caching optimization.\nI will abandon this patch.","commit_id":"a8d4fa29ddf0b9fd4c520bbe837f3bc3ff451533"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"f5ba67164d7d8fb74486f87e8947734c75a3d260","unresolved":true,"context_lines":[{"line_number":10,"context_line":"after each action completion, resulting in O(n^2+) complexity"},{"line_number":11,"context_line":"for large action plans."},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"This patch adds a new \u0027flow_type\u0027 config option to"},{"line_number":14,"context_line":"DefaultWorkFlowEngine. When set to \u0027linear\u0027, actions are"},{"line_number":15,"context_line":"topologically sorted using Kahn\u0027s algorithm in O(V+E), then"},{"line_number":16,"context_line":"executed via a linear_flow. This preserves all Taskflow"},{"line_number":17,"context_line":"revert/abort/notification mechanisms while avoiding the"},{"line_number":18,"context_line":"graph re-analysis overhead."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"The default value remains \u0027graph\u0027 for backward compatibility."},{"line_number":21,"context_line":""},{"line_number":22,"context_line":"Note: \u0027linear\u0027 mode does not support the ANY execution rule,"},{"line_number":23,"context_line":"which requires graph_flow dependency links."},{"line_number":24,"context_line":""},{"line_number":25,"context_line":"Closes-Bug: #2144961"},{"line_number":26,"context_line":"Change-Id: I73a4a06767872601a0c8b051e6632d04c3df2b9e"},{"line_number":27,"context_line":"Signed-off-by: hasanacar \u003chasan.acar@tubitak.gov.tr\u003e"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"d37cf433_151440ac","line":25,"range":{"start_line":13,"start_character":0,"end_line":25,"end_character":20},"updated":"2026-03-24 13:32:45.000000000","message":"this is a fundamental change to the mental model of how watcher works and its probably not something we shoudl allow to be configured externally.\nthat woudl create a large oeprational and maintance burden in unserstanidng which stagies wrok with which flow_type and requrie much much more testing then is proposed in this patch.\n\nits really a feature request not a bug and one i think need proper discussion and design before we move forward to a implementation.","commit_id":"a8d4fa29ddf0b9fd4c520bbe837f3bc3ff451533"}]}