Oregami
Repositories/oxedyne/fe2o3

fe2o3_austenite/tests/watch.rs

34 operations, oldest first

created by r1870400018:68736, which is this file's identity for as long as the history lasts, whatever it is later renamed to

read the file · who wrote it · compare two states

An operation carries no clock of its own. Each row is therefore dated by the first mark written after it by the same replica, which is one copy of the repository, and that mark is where the command that wrote the operation ended.

+ signed and verified, ? signed by a key this forge does not carry, - unsigned

OperationAuthorWhatWhen
+r1870400018:68736replica 1870400018created fe2o3_austenite/tests/watch.rs43 days ago under "fe2o3_austenite eval: a closure keeps its fingerprint, so a capture reached by many paths is hashed once (fix: eval_survey fletcher spun for 20 min; red: a 48-deep chain of shared closures timed out at 20 s, 0.01 s after)"
+r1870400018:68737replica 1870400018inserted 24.8 KiB43 days ago under "fe2o3_austenite eval: a closure keeps its fingerprint, so a capture reached by many paths is hashed once (fix: eval_survey fletcher spun for 20 min; red: a 48-deep chain of shared closures timed out at 20 s, 0.01 s after)"
+r1870400018:69572replica 1870400018inserted 7 bytes42 days ago under "fe2o3_austenite eval: a closure captures the names it mentions as variables, not the field of a field access or the key of a named argument (warm-pass fix)"
+r1870400018:69573replica 1870400018replaced 566 bytes with 1.9 KiB42 days ago under "fe2o3_austenite eval: a closure captures the names it mentions as variables, not the field of a field access or the key of a named argument (warm-pass fix)"
+r1870400018:69574replica 1870400018replaced 351 bytes with 1.8 KiB42 days ago under "fe2o3_austenite eval: a closure captures the names it mentions as variables, not the field of a field access or the key of a named argument (warm-pass fix)"
+r1870400018:69575replica 1870400018replaced 5 bytes with 11 bytes42 days ago under "fe2o3_austenite eval: a closure captures the names it mentions as variables, not the field of a field access or the key of a named argument (warm-pass fix)"
+r1870400018:69576replica 1870400018replaced 5 bytes with 11 bytes42 days ago under "fe2o3_austenite eval: a closure captures the names it mentions as variables, not the field of a field access or the key of a named argument (warm-pass fix)"
+r1870400018:69577replica 1870400018replaced 5 bytes with 11 bytes42 days ago under "fe2o3_austenite eval: a closure captures the names it mentions as variables, not the field of a field access or the key of a named argument (warm-pass fix)"
+r1870400018:69578replica 1870400018replaced 5 bytes with 11 bytes42 days ago under "fe2o3_austenite eval: a closure captures the names it mentions as variables, not the field of a field access or the key of a named argument (warm-pass fix)"
+r1870400018:69579replica 1870400018replaced 5 bytes with 11 bytes42 days ago under "fe2o3_austenite eval: a closure captures the names it mentions as variables, not the field of a field access or the key of a named argument (warm-pass fix)"
+r1870400018:69580replica 1870400018inserted 1.1 KiB42 days ago under "fe2o3_austenite eval: a closure captures the names it mentions as variables, not the field of a field access or the key of a named argument (warm-pass fix)"
+r1870400018:69581replica 1870400018replaced 5 bytes with 11 bytes42 days ago under "fe2o3_austenite eval: a closure captures the names it mentions as variables, not the field of a field access or the key of a named argument (warm-pass fix)"
+r1870400018:69582replica 1870400018replaced 5 bytes with 11 bytes42 days ago under "fe2o3_austenite eval: a closure captures the names it mentions as variables, not the field of a field access or the key of a named argument (warm-pass fix)"
+r1870400018:69583replica 1870400018replaced 10 bytes with 16 bytes42 days ago under "fe2o3_austenite eval: a closure captures the names it mentions as variables, not the field of a field access or the key of a named argument (warm-pass fix)"
+r1870400018:69584replica 1870400018inserted 4.5 KiB42 days ago under "fe2o3_austenite eval: a closure captures the names it mentions as variables, not the field of a field access or the key of a named argument (warm-pass fix)"
+r1870400018:69787replica 1870400018replaced 10 bytes with 8 bytes42 days ago under "fe2o3_o3db_sync close: the server bots are the first stage of a close, so a chunked put accepted before close() is stored or refused, never told Written and lost (A3 QA M2), and a failure to send a store's next step reaches its caller (m1)"
+r1870400018:69788replica 1870400018inserted 2.7 KiB42 days ago under "fe2o3_o3db_sync close: the server bots are the first stage of a close, so a chunked put accepted before close() is stored or refused, never told Written and lost (A3 QA M2), and a failure to send a store's next step reaches its caller (m1)"
+r1870400018:69789replica 1870400018replaced 21 bytes with 829 bytes42 days ago under "fe2o3_o3db_sync close: the server bots are the first stage of a close, so a chunked put accepted before close() is stored or refused, never told Written and lost (A3 QA M2), and a failure to send a store's next step reaches its caller (m1)"
+r1870400018:69790replica 1870400018replaced 85 bytes with 4.7 KiB42 days ago under "fe2o3_o3db_sync close: the server bots are the first stage of a close, so a chunked put accepted before close() is stored or refused, never told Written and lost (A3 QA M2), and a failure to send a store's next step reaches its caller (m1)"
+r1870400018:74625replica 1870400018inserted 5 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74626replica 1870400018inserted 5 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74627replica 1870400018inserted 5 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74628replica 1870400018inserted 5 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74629replica 1870400018inserted 5 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74630replica 1870400018inserted 5 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74631replica 1870400018inserted 5 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74632replica 1870400018inserted 5 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74633replica 1870400018inserted 5 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74634replica 1870400018inserted 5 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74635replica 1870400018inserted 24 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74636replica 1870400018inserted 5 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74637replica 1870400018inserted 12 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74638replica 1870400018inserted 5 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"
+r1870400018:74639replica 1870400018inserted 24 bytes33 days ago under "fe2o3_austenite: --input key=value reaches the document as sys.inputs, as typst compile --input gives it (D-20261018-03 U5) -- the library was one thread-local scope with an empty sys.inputs and no route to fill it. World now carries its library (Typst's World::library), built by eval::library_with_inputs: the shared scope when there are no inputs, else a copy whose sys module (foundations::sys_module) holds them as strings; eval_module, eval_string and call_closure evaluate against engine.world.library. Session::set_inputs folds a repeated key as Typst's dictionary does (first place, last value) and each compile builds its world with them; caches::configuration now hashes the inputs in the order given, since keys() and a for loop read that order (the order-is-not-a-change assertion in eval_session is turned round), and caches::sys_inputs, which read the empty library, is gone. compile::parse_input splits at the first =, trims key and value, and refuses a missing = or an empty key, naming the argument. CLI: --input on plain compiles, --explain, --watch (Bare), austenite build and austenite watch (Spec.inputs, kept across a settings reload); the curated reader refuses it by name. Adds austenite --version (crate version and engine commit). Tests tests/sys_inputs.rs (9: empty dictionary, one, several with a repeated key, .at with a default, sys.version unchanged, a session's new inputs move the configuration, the parse, watch/bare/build) and tests/sys_inputs_cli.rs (7: plain, --explain, build, none, the control, malformed refused, --version); red before (the installed binary read --input as the source and sys.inputs.at(\"engine\", default: none) gave none). Package suite green but for oracle corpus_roots_compile_and_match_the_typst_oracle (external corpus drift: oxeweb-techspec headings, float-fixture, baselines moved); Hematite guide 395 pages before and after"