Prepare for schemars v1.0.0

.gitattributes

@@ -0,0 +1 @@
+crates/numcodecs-wasm-host-reproducible/tests/round.wasm filter=lfs diff=lfs merge=lfs -text

.github/workflows/ci.yml

@@ -114,6 +114,8 @@ jobs:
     steps:
       - name: Checkout the Repository
         uses: actions/checkout@v2
+        with:
+          lfs: true
 
       - name: Install the Rust toolchain
         uses: actions-rs/toolchain@v1

Cargo.toml

@@ -81,7 +81,7 @@ burn = { version = "0.17", default-features = false }
 clap = { version = "4.5", default-features = false }
 convert_case = { version = "0.8", default-features = false }
 format_serde_error = { version = "0.3", default-features = false }
-indexmap = { version = "2.7.1", default-features = false }
+indexmap = { version = "2.10", default-features = false }
 itertools = { version = "0.14", default-features = false }
 log = { version = "0.4.27", default-features = false }
 simple_logger = { version = "5.0", default-features = false }
@@ -100,7 +100,7 @@ pyo3-error = { version = "0.5", default-features = false }
 pyo3-log = { version = "0.12.4", default-features = false }
 pythonize = { version = "0.25", default-features = false }
 rand = { version = "0.9.1", default-features = false }
-schemars = { version = "=1.0.0-alpha.15", default-features = false }
+schemars = { version = "1.0.3", default-features = false }
 scratch = { version = "1.0", default-features = false }
 semver = { version = "1.0.23", default-features = false }
 serde = { version = "1.0.218", default-features = false }
@@ -122,7 +122,7 @@ wasmtime = { version = "33.0", default-features = false }
 wasmtime_runtime_layer = { version = "33.0", default-features = false }
 wasm-encoder = { version = "0.235", default-features = false }
 wasm_runtime_layer = { version = "0.5", default-features = false }
-wit-bindgen = { version = "0.42", default-features = false }
+wit-bindgen = { version = "0.43", default-features = false }
 wit-component = { version = "0.235", default-features = false }
 wit-parser = { version = "0.235", default-features = false }
 wyhash = { version = "0.6", default-features = false }

codecs/fourier-network/tests/schema.json

@@ -37,7 +37,7 @@
       ],
       "format": "uint",
       "minimum": 1,
-      "description": "The optional mini-batch size used during training\n\n Setting the mini-batch size to `None` disables the use of batching,\n i.e. the network is trained using one large batch that includes the\n full data."
+      "description": "The optional mini-batch size used during training\n\nSetting the mini-batch size to `None` disables the use of batching,\ni.e. the network is trained using one large batch that includes the\nfull data."
     },
     "seed": {
       "type": "integer",
@@ -61,7 +61,7 @@
     "mini_batch_size",
     "seed"
   ],
-  "description": "Fourier network codec which trains and overfits a fourier feature neural\n network on encoding and predicts during decoding.\n\n The approach is based on the papers by Tancik et al. 2020\n (<https://dl.acm.org/doi/abs/10.5555/3495724.3496356>)\n and by Huang and Hoefler 2020 (<https://arxiv.org/abs/2210.12538>).",
+  "description": "Fourier network codec which trains and overfits a fourier feature neural\nnetwork on encoding and predicts during decoding.\n\nThe approach is based on the papers by Tancik et al. 2020\n(<https://dl.acm.org/doi/abs/10.5555/3495724.3496356>)\nand by Huang and Hoefler 2020 (<https://arxiv.org/abs/2210.12538>).",
   "title": "FourierNetworkCodec",
   "$schema": "https://json-schema.org/draft/2020-12/schema"
 }

codecs/jpeg2000/src/ffi/image.rs

@@ -25,8 +25,9 @@ impl Image {
     ) -> Result<Self, Jpeg2000Error> {
         let mut image = std::ptr::null_mut();
 
-        if unsafe { openjpeg_sys::opj_read_header(stream.as_raw(), decoder.as_raw(), &mut image) }
-            != 1
+        if unsafe {
+            openjpeg_sys::opj_read_header(stream.as_raw(), decoder.as_raw(), &raw mut image)
+        } != 1
         {
             return Err(Jpeg2000Error::InvalidMainHeader);
         }
@@ -56,7 +57,7 @@ impl Image {
         let image = NonNull::new(unsafe {
             openjpeg_sys::opj_image_create(
                 1,
-                &mut image_params,
+                &raw mut image_params,
                 openjpeg_sys::OPJ_COLOR_SPACE::OPJ_CLRSPC_GRAY,
             )
         })

codecs/jpeg2000/src/ffi/mod.rs

@@ -103,7 +103,7 @@ pub fn encode_into<T: Jpeg2000Element>(
     let mut image = Image::from_gray_data(data, width, height)?;
 
     if unsafe {
-        openjpeg_sys::opj_setup_encoder(encoder.as_raw(), &mut encode_params, image.as_raw())
+        openjpeg_sys::opj_setup_encoder(encoder.as_raw(), &raw mut encode_params, image.as_raw())
     } != 1
     {
         return Err(Jpeg2000Error::EncoderSetupError);
@@ -136,7 +136,7 @@ pub fn decode<T: Jpeg2000Element>(bytes: &[u8]) -> Result<(Vec<T>, (usize, usize
     let mut decode_params = unsafe { decode_params.assume_init() };
     decode_params.decod_format = 1; // JP2
 
-    if unsafe { openjpeg_sys::opj_setup_decoder(decoder.as_raw(), &mut decode_params) } != 1 {
+    if unsafe { openjpeg_sys::opj_setup_decoder(decoder.as_raw(), &raw mut decode_params) } != 1 {
         return Err(Jpeg2000Error::DecoderSetupError);
     }
 

codecs/jpeg2000/tests/schema.json

@@ -54,7 +54,7 @@
       "description": "Lossless compression"
     }
   ],
-  "description": "Codec providing compression using JPEG 2000.\n\n Arrays that are higher-dimensional than 2D are encoded by compressing each\n 2D slice with JPEG 2000 independently. Specifically, the array's shape is\n interpreted as `[.., height, width]`. If you want to compress 2D slices\n along two different axes, you can swizzle the array axes beforehand.",
+  "description": "Codec providing compression using JPEG 2000.\n\nArrays that are higher-dimensional than 2D are encoded by compressing each\n2D slice with JPEG 2000 independently. Specifically, the array's shape is\ninterpreted as `[.., height, width]`. If you want to compress 2D slices\nalong two different axes, you can swizzle the array axes beforehand.",
   "properties": {
     "_version": {
       "type": "string",

codecs/pco/tests/schema.json

@@ -21,7 +21,7 @@
         11,
         12
       ],
-      "description": "Compression level, ranging from 0 (weak) over 8 (very good) to 12\n (expensive)"
+      "description": "Compression level, ranging from 0 (weak) over 8 (very good) to 12\n(expensive)"
     },
     "_version": {
       "type": "string",
@@ -46,7 +46,7 @@
           "required": [
             "mode"
           ],
-          "description": "Automatically detects a good mode.\n\n This works well most of the time, but costs some compression time and\n can select a bad mode in adversarial cases."
+          "description": "Automatically detects a good mode.\n\nThis works well most of the time, but costs some compression time and\ncan select a bad mode in adversarial cases."
         },
         {
           "type": "object",
@@ -63,7 +63,7 @@
         },
         {
           "type": "object",
-          "description": "Tries using the `FloatMult` mode with a given base.\n\n Only applies to floating-point types.",
+          "description": "Tries using the `FloatMult` mode with a given base.\n\nOnly applies to floating-point types.",
           "properties": {
             "float_mult_base": {
               "type": "number",
@@ -82,7 +82,7 @@
         },
         {
           "type": "object",
-          "description": "Tries using the `FloatQuant` mode with the given number of bits of\n quantization.\n\n Only applies to floating-point types.",
+          "description": "Tries using the `FloatQuant` mode with the given number of bits of\nquantization.\n\nOnly applies to floating-point types.",
           "properties": {
             "float_quant_bits": {
               "type": "integer",
@@ -102,7 +102,7 @@
         },
         {
           "type": "object",
-          "description": "Tries using the `IntMult` mode with a given base.\n\n Only applies to integer types.",
+          "description": "Tries using the `IntMult` mode with a given base.\n\nOnly applies to integer types.",
           "properties": {
             "int_mult_base": {
               "type": "integer",
@@ -135,7 +135,7 @@
           "required": [
             "delta"
           ],
-          "description": "Automatically detects a detects a good delta encoding.\n\n This works well most of the time, but costs some compression time and\n can select a bad delta encoding in adversarial cases."
+          "description": "Automatically detects a detects a good delta encoding.\n\nThis works well most of the time, but costs some compression time and\ncan select a bad delta encoding in adversarial cases."
         },
         {
           "type": "object",
@@ -148,11 +148,11 @@
           "required": [
             "delta"
           ],
-          "description": "Never uses delta encoding.\n\n This is best if your data is in a random order or adjacent numbers have\n no relation to each other."
+          "description": "Never uses delta encoding.\n\nThis is best if your data is in a random order or adjacent numbers have\nno relation to each other."
         },
         {
           "type": "object",
-          "description": "Tries taking nth order consecutive deltas.\n\n Supports a delta encoding order up to 7. For instance, 1st order is\n just regular delta encoding, 2nd is deltas-of-deltas, etc. It is legal\n to use 0th order, but it is identical to None.",
+          "description": "Tries taking nth order consecutive deltas.\n\nSupports a delta encoding order up to 7. For instance, 1st order is\njust regular delta encoding, 2nd is deltas-of-deltas, etc. It is legal\nto use 0th order, but it is identical to None.",
           "properties": {
             "delta_encoding_order": {
               "type": "integer",
@@ -189,15 +189,15 @@
           "required": [
             "delta"
           ],
-          "description": "Tries delta encoding according to an extra latent variable of\n \"lookback\".\n\n This can improve compression ratio when there are nontrivial patterns\n in the array, but reduces compression speed substantially."
+          "description": "Tries delta encoding according to an extra latent variable of\n\"lookback\".\n\nThis can improve compression ratio when there are nontrivial patterns\nin the array, but reduces compression speed substantially."
         }
       ]
     }
   ],
   "oneOf": [
     {
       "type": "object",
-      "description": "Divide the chunk into equal pages of up to this many numbers.\n\n For example, with equal pages up to 100,000, a chunk of 150,000 numbers\n would be divided into 2 pages, each of 75,000 numbers.",
+      "description": "Divide the chunk into equal pages of up to this many numbers.\n\nFor example, with equal pages up to 100,000, a chunk of 150,000 numbers\nwould be divided into 2 pages, each of 75,000 numbers.",
       "properties": {
         "equal_pages_up_to": {
           "type": "integer",

codecs/random-projection/tests/schema.json

@@ -17,7 +17,7 @@
   "required": [
     "seed"
   ],
-  "description": "Codec that uses random projections to reduce the dimensionality of high-\n dimensional data to compress it.\n\n A two-dimensional array of shape `$N \\times D$` is encoded as n array of\n shape `$N \\times K$`, where `$K$` is either set explicitly or chosen using\n the the Johnson-Lindenstrauss lemma. For `$K$` to be smaller than `$D$`,\n `$D$` must be quite large. Therefore, this codec should only applied on\n large datasets as it otherwise significantly inflates the data size instead\n of reducing it.\n\n Choosing a lower distortion rate `epsilon` will improve the quality of the\n lossy compression, i.e. reduce the compression error, at the cost of\n increasing `$K$`.\n\n This codec only supports finite floating point data.",
+  "description": "Codec that uses random projections to reduce the dimensionality of high-\ndimensional data to compress it.\n\nA two-dimensional array of shape `$N \\times D$` is encoded as n array of\nshape `$N \\times K$`, where `$K$` is either set explicitly or chosen using\nthe the Johnson-Lindenstrauss lemma. For `$K$` to be smaller than `$D$`,\n`$D$` must be quite large. Therefore, this codec should only applied on\nlarge datasets as it otherwise significantly inflates the data size instead\nof reducing it.\n\nChoosing a lower distortion rate `epsilon` will improve the quality of the\nlossy compression, i.e. reduce the compression error, at the cost of\nincreasing `$K$`.\n\nThis codec only supports finite floating point data.",
   "allOf": [
     {
       "oneOf": [
@@ -39,7 +39,7 @@
             "reduction",
             "epsilon"
           ],
-          "description": "The reduced dimensionality `$K$` is derived from `epsilon`, as defined\n by the Johnson-Lindenstrauss lemma."
+          "description": "The reduced dimensionality `$K$` is derived from `epsilon`, as defined\nby the Johnson-Lindenstrauss lemma."
         },
         {
           "type": "object",
@@ -59,7 +59,7 @@
             "reduction",
             "k"
           ],
-          "description": "The reduced dimensionality `$K$`, to which the data is projected, is\n given explicitly."
+          "description": "The reduced dimensionality `$K$`, to which the data is projected, is\ngiven explicitly."
         }
       ]
     },
@@ -76,7 +76,7 @@
           "required": [
             "projection"
           ],
-          "description": "The random projection matrix is dense and its components are sampled\n from `$\\text{N}\\left( 0, \\frac{1}{k} \\right)$`"
+          "description": "The random projection matrix is dense and its components are sampled\nfrom `$\\text{N}\\left( 0, \\frac{1}{k} \\right)$`"
         },
         {
           "type": "object",
@@ -88,7 +88,7 @@
               ],
               "exclusiveMinimum": 0.0,
               "maximum": 1.0,
-              "description": "The `density` of the sparse projection matrix.\n\n Setting `density` to `$\\frac{1}{3}$` reproduces the settings by\n Achlioptas [^1]. If `density` is `None`, it is set to\n `$\\frac{1}{\\sqrt{d}}$`,\n the minimum density as recommended by Li et al [^2].\n\n\n [^1]: Achlioptas, D. (2003). Database-friendly random projections:\n       Johnson-Lindenstrauss with binary coins. *Journal of Computer\n       and System Sciences*, 66(4), 671-687. Available from:\n       [doi:10.1016/S0022-0000(03)00025-4](https://doi.org/10.1016/S0022-0000(03)00025-4).\n\n [^2]: Li, P., Hastie, T. J., and Church, K. W. (2006). Very sparse\n       random projections. In *Proceedings of the 12th ACM SIGKDD\n       international conference on Knowledge discovery and data\n       mining (KDD '06)*. Association for Computing Machinery, New\n       York, NY, USA, 287–296. Available from:\n       [doi:10.1145/1150402.1150436](https://doi.org/10.1145/1150402.1150436)."
+              "description": "The `density` of the sparse projection matrix.\n\nSetting `density` to `$\\frac{1}{3}$` reproduces the settings by\nAchlioptas [^1]. If `density` is `None`, it is set to\n`$\\frac{1}{\\sqrt{d}}$`,\nthe minimum density as recommended by Li et al [^2].\n\n\n[^1]: Achlioptas, D. (2003). Database-friendly random projections:\n      Johnson-Lindenstrauss with binary coins. *Journal of Computer\n      and System Sciences*, 66(4), 671-687. Available from:\n      [doi:10.1016/S0022-0000(03)00025-4](https://doi.org/10.1016/S0022-0000(03)00025-4).\n\n[^2]: Li, P., Hastie, T. J., and Church, K. W. (2006). Very sparse\n      random projections. In *Proceedings of the 12th ACM SIGKDD\n      international conference on Knowledge discovery and data\n      mining (KDD '06)*. Association for Computing Machinery, New\n      York, NY, USA, 287–296. Available from:\n      [doi:10.1145/1150402.1150436](https://doi.org/10.1145/1150402.1150436)."
             },
             "projection": {
               "type": "string",
@@ -98,7 +98,7 @@
           "required": [
             "projection"
           ],
-          "description": "The random projection matrix is sparse where only `density`% of entries\n are non-zero.\n\n The matrix's components are sampled from\n\n - `$-\\sqrt{\\frac{1}{k \\cdot density}}$` with probability\n   `$0.5 \\cdot density$`\n - `$0$` with probability `$1 - density$`\n - `$+\\sqrt{\\frac{1}{k \\cdot density}}$` with probability\n   `$0.5 \\cdot density$`"
+          "description": "The random projection matrix is sparse where only `density`% of entries\nare non-zero.\n\nThe matrix's components are sampled from\n\n- `$-\\sqrt{\\frac{1}{k \\cdot density}}$` with probability\n  `$0.5 \\cdot density$`\n- `$0$` with probability `$1 - density$`\n- `$+\\sqrt{\\frac{1}{k \\cdot density}}$` with probability\n  `$0.5 \\cdot density$`"
         }
       ]
     }

codecs/sperr/tests/schema.json

@@ -60,7 +60,7 @@
       "description": "Fixed point-wise (absolute) error"
     }
   ],
-  "description": "Codec providing compression using SPERR.\n\n Arrays that are higher-dimensional than 3D are encoded by compressing each\n 3D slice with SPERR independently. Specifically, the array's shape is\n interpreted as `[.., depth, height, width]`. If you want to compress 3D\n slices along three different axes, you can swizzle the array axes\n beforehand.",
+  "description": "Codec providing compression using SPERR.\n\nArrays that are higher-dimensional than 3D are encoded by compressing each\n3D slice with SPERR independently. Specifically, the array's shape is\ninterpreted as `[.., depth, height, width]`. If you want to compress 3D\nslices along three different axes, you can swizzle the array axes\nbeforehand.",
   "properties": {
     "_version": {
       "type": "string",

codecs/sz3/tests/schema.json

@@ -121,7 +121,7 @@
   "oneOf": [
     {
       "type": "object",
-      "description": "Errors are bounded by *both* the absolute and relative error, i.e. by\n whichever bound is stricter",
+      "description": "Errors are bounded by *both* the absolute and relative error, i.e. by\nwhichever bound is stricter",
       "properties": {
         "eb_abs": {
           "type": "number",
@@ -146,7 +146,7 @@
     },
     {
       "type": "object",
-      "description": "Errors are bounded by *either* the absolute or relative error, i.e. by\n whichever bound is weaker",
+      "description": "Errors are bounded by *either* the absolute or relative error, i.e. by\nwhichever bound is weaker",
       "properties": {
         "eb_abs": {
           "type": "number",