It is currently not possible to have a multi-layer (e.g.) transform where one of the data products comes from an unfold. For example, consider the following minimal reproducer of a provider + unfold + multi-layer transform:
experimental::layer_generator gen;
gen.add_layer("event", {"job", 2u});
experimental::framework_graph g{driver_for_test(gen)};
// Provide maximum number to be used by 'iota' unfold
g.provide("provide_max_number", provide_max_number, concurrency::unlimited)
.output_product(product_query{.creator = "input", .layer = "event", .suffix = "max_number"});
// Unfold from 0 to that maximum number (exclusive)
g.unfold<iota>("iota", &iota::predicate, &iota::unfold, concurrency::unlimited, "subevent")
.input_family(product_query{.creator = "input", .layer = "event", .suffix = "max_number"})
.output_product_suffixes("new_number");
// Read the new_number from the unfold and max_number from the event
g.transform("add_max_and_new", [](unsigned int i, unsigned int j) { return i + j; }, concurrency::unlimited)
.input_family(
product_query{.creator = "iota", .layer = "subevent", .suffix = "new_number"},
product_query{.creator = "input", .layer = "event", .suffix = "max_number"})
.output_product_suffixes("result");
g.execute();
// event 0: max_number=10, new_number in [0,9] -> 10 executions
// event 1: max_number=20, new_number in [0,19] -> 20 executions
CHECK(g.execution_count("iota") == 2u;
CHECK(g.execution_count("add_max_and_new") == 30u);When running this job, one sees:
/Users/kyleknoepfel/work/phlex-devel/srcs/phlex/test/unfold.cpp:184: FAILED:
CHECK( g.execution_count("add_max_and_new") == 30u )
with expansion:
0 == 30
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/event] Cached messages: 2
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/event] Product for [event:1]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/event] Product for [event:0]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Cached messages: 30
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:16]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:4]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:2]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:3]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:0, subevent:0]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:10]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:0]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:0, subevent:1]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:0, subevent:5]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:7]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:6]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:0, subevent:3]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:13]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:0, subevent:9]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:14]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:9]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:0, subevent:8]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:0, subevent:4]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:0, subevent:2]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:17]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:5]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:8]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:1]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:0, subevent:7]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:18]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:15]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:11]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:19]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:0, subevent:6]
[2026-04-23 11:54:31.581] [warning] [add_max_and_new/subevent] Product for [event:1, subevent:12]
[2026-04-23 11:54:31.581] [info]
Seen layers:
job
│
└ event: 2
│
└ subevent: 30
It is currently not possible to have a multi-layer (e.g.) transform where one of the data products comes from an unfold. For example, consider the following minimal reproducer of a provider + unfold + multi-layer transform:
When running this job, one sees: