| 77 | implicit unsigned 32-bit integer | No tag |
@@ -299,7 +322,7 @@ Some things to note.
2. For an untagged positive integer, the constant is treated as of the smallest type into which it fits (i.e. 8,16,32, or 64 bit).
3. For signed byte and short, the value is sign extended to 32 bits and then treated as an unsigned int value, but maintaining the bit-pattern.
4. For double, and signed|unsigned long long, they are converted as specified in the section on parameter encode/decode.
-5. In order to support mutiple filters, the argument to ''_Filter'' may be a pipeline separated (using '|') to specify a list of filters specs.
+5. In order to support mutiple filters, the argument to *\_Filter* may be a pipeline separated (using '|') to specify a list of filters specs.
# Dynamic Loading Process {#filters_Process}
@@ -314,7 +337,7 @@ The default directory is:
* "/usr/local/hdf5/lib/plugin” for linux/unix operating systems (including Cygwin)
* “%ALLUSERSPROFILE%\\hdf5\\lib\\plugin” for Windows systems, although the code does not appear to explicitly use this path.
-The default may be overridden using the environment variable __HDF5_PLUGIN_PATH__.
+The default may be overridden using the environment variable *HDF5\_PLUGIN\_PATH*.
## Plugin Library Naming {#filters_Pluginlib}
@@ -332,10 +355,10 @@ Given a plugin directory, HDF5 examines every file in that directory that confor
For each dynamic library located using the previous patterns, HDF5 attempts to load the library and attempts to obtain information from it.
Specifically, It looks for two functions with the following signatures.
-1. __H5PL_type_t H5PLget_plugin_type(void)__ — This function is expected to return the constant value __H5PL_TYPE_FILTER__ to indicate that this is a filter library.
-2. __const void* H5PLget_plugin_info(void)__ — This function returns a pointer to a table of type __H5Z_class2_t__.
+1. *H5PL\_type\_t H5PLget\_plugin\_type(void)* — This function is expected to return the constant value *H5PL\_TYPE\_FILTER* to indicate that this is a filter library.
+2. *const void* H5PLget\_plugin\_info(void)* — This function returns a pointer to a table of type *H5Z\_class2\_t*.
This table contains the necessary information needed to utilize the filter both for reading and for writing.
- In particular, it specifies the filter id implemented by the library and it must match that id specified for the variable in __nc_def_var_filter__ in order to be used.
+ In particular, it specifies the filter id implemented by the library and it must match that id specified for the variable in *nc\_def\_var\_filter* in order to be used.
If plugin verification fails, then that plugin is ignored and the search continues for another, matching plugin.
@@ -346,9 +369,9 @@ For Zarr, filters are represented using the JSON notation.
Each filter is defined by a JSON dictionary, and each such filter dictionary
is guaranteed to have a key named "id" whose value is a unique string defining the filter algorithm: "lz4" or "bzip2", for example.
-The parameters of the filter are defined by additional -- algorithm specific -- keys in the filter dictionary.
+The parameters of the filter are defined by additional — algorithm specific — keys in the filter dictionary.
One commonly used filter is "blosc", which has a JSON dictionary of this form.
-````
+````
{
"id": "blosc",
"cname": "lz4",
@@ -358,9 +381,9 @@ One commonly used filter is "blosc", which has a JSON dictionary of this form.
````
So it has three parameters:
-1. "cname" -- the sub-algorithm used by the blosc compressor, LZ4 in this case.
-2. "clevel" -- the compression level, 5 in this case.
-3. "shuffle" -- is the input shuffled before compression, yes (1) in this case.
+1. "cname" — the sub-algorithm used by the blosc compressor, LZ4 in this case.
+2. "clevel" — the compression level, 5 in this case.
+3. "shuffle" — is the input shuffled before compression, yes (1) in this case.
NCZarr has four constraints that must be met.
@@ -371,7 +394,7 @@ This means that some mechanism is needed to translate between the HDF5 id+parame
3. It must be possible to modify the set of visible parameters in response to environment information such as the type of the associated variable; this is required to mimic the corresponding HDF5 capability.
4. It must be possible to use filters even if HDF5 support is disabled.
-Note that the term "visible parameters" is used here to refer to the parameters provided by "nc_def_var_filter" or those stored in the dataset's metadata as provided by the JSON codec. The term "working parameters" refers to the parameters given to the compressor itself and derived from the visible parameters.
+Note that the term "visible parameters" is used here to refer to the parameters provided by "nc\_def\_var\_filter" or those stored in the dataset's metadata as provided by the JSON codec. The term "working parameters" refers to the parameters given to the compressor itself and derived from the visible parameters.
The standard authority for defining Zarr filters is the list supported by the NumCodecs project [7].
Comparing the set of standard filters (aka codecs) defined by NumCodecs to the set of standard filters defined by HDF5 [3], it can be seen that the two sets overlap, but each has filters not defined by the other.
@@ -382,25 +405,27 @@ Rather, it is preferable for there be some extensible way to associate the JSON
The mechanism provided to address these issues is similar to that taken by HDF5.
A shared library must exist that has certain well-defined entry points that allow the NCZarr code to determine information about a Codec.
The shared library exports a well-known function name to access Codec information and relate it to a corresponding HDF5 implementation,
+Note that the shared library may optionally be the same library containing the HDF5
+filter processor.
## Processing Overview
There are several paths by which the NCZarr filter API is invoked.
-1. The nc_def_var_filter function is invoked on a variable or
+1. The nc\_def\_var\_filter function is invoked on a variable or
(1a) the metadata for a variable is read when opening an existing variable that has associated Codecs.
2. The visible parameters are converted to a set of working parameters.
3. The filter is invoked with the working parameters.
4. The dataset is closed using the final set of visible parameters.
-### Step 1: Invoking nc_def_var_filter
+### Step 1: Invoking nc\_def\_var\_filter
-In this case, the filter plugin is located and the set of visible parameters (from nc_def_var_filter) are provided.
+In this case, the filter plugin is located and the set of visible parameters (from nc\_def\_var\_filter) are provided.
### Step 1a: Reading metadata
In this case, the codec is read from the metadata and must be converted to a visible set of HDF5 style parameters.
-It is possible that this set of visible parameters differs from the set that was provided by nc_def_var_filter.
+It is possible that this set of visible parameters differs from the set that was provided by nc\_def\_var\_filter.
If this is important, then the filter implementation is responsible for marking this difference using, for example, different number of parameters or some differing value.
### Step 2: Convert visible parameters to working parameters
@@ -423,30 +448,30 @@ If no change is detected, then re-writing the compressor metadata may be avoided
Currently, there is no way to specify use of a filter via Codec through
the netcdf-c API. Rather, one must know the HDF5 id and parameters of
-the filter of interest and use the functions ''nc_def_var_filter'' and ''nc_inq_var_filter''.
+the filter of interest and use the functions *nc\_def\_var\_filter* and *nc\_inq\_var\_filter*.
Internally, the NCZarr code will use information about known Codecs to convert the HDF5 filter reference to the corresponding Codec.
-This restriction also holds for the specification of filters in ''ncgen'' and ''nccopy''.
+This restriction also holds for the specification of filters in *ncgen* and *nccopy*.
This limitation may be lifted in the future.
## Special Codecs Attribute
-A new special attribute is defined called ''_Codecs'' in parallel to the current ''_Filters'' special attribute. Its value is a string containing the JSON representation of the Codecs associated with a given variable.
+A new special attribute is defined called *\_Codecs* in parallel to the current *\_Filters* special attribute. Its value is a string containing the JSON representation of the Codecs associated with a given variable.
This can be especially useful when a file is unreadable because it uses a filter not available to the netcdf-c library.
-That is, no implementation was found in the e.g. ''HDF5_PLUGIN_PATH'' directory.
-In this case ''ncdump -hs'' will display the raw Codec information so that it may be possible to see what filter is missing.
+That is, no implementation was found in the e.g. *HDF5\_PLUGIN\_PATH* directory.
+In this case *ncdump -hs* will display the raw Codec information so that it may be possible to see what filter is missing.
## Pre-Processing Filter Libraries
The process for using filters for NCZarr is defined to operate in several steps.
First, as with HDF5, all shared libraries in a specified directory
-(e.g. ''HDF5_PLUGIN_PATH'') are scanned.
+(e.g. *HDF5\_PLUGIN\_PATH*) are scanned.
They are interrogated to see what kind of library they implement, if any.
This interrogation operates by seeing if certain well-known (function) names are defined in this library.
There will be two library types:
-1. HDF5 -- exports a specific API: "H5Z\_plugin\_type" and "H5Z\_get\_plugin\_info".
-2. Codec -- exports a specific API: "NCZ\_get\_codec\_info"
+1. HDF5 — exports a specific API: "H5Z\_plugin\_type" and "H5Z\_get\_plugin\_info".
+2. Codec — exports a specific API: "NCZ\_get\_codec\_info"
Note that a given library can export either or both of these APIs.
This means that we can have three types of libraries:
@@ -455,7 +480,7 @@ This means that we can have three types of libraries:
2. Codec only
3. HDF5 + Codec
-Suppose that our ''HDF5_PLUGIN_PATH'' location has an HDF5-only library.
+Suppose that our *HDF5\_PLUGIN\_PATH* location has an HDF5-only library.
Then by adding a corresponding, separate, Codec-only library to that same location, it is possible to make an HDF5 library usable by NCZarr.
It is possible to do this without having to modify the HDF5-only library.
Over time, it is possible to merge an HDF5-only library with a Codec-only library to produce a single, combined library.
@@ -466,7 +491,7 @@ The netcdf-c library processes all of the shared libraries by interrogating each
Any libraries that do not export one or both of the well-known APIs is ignored.
Internally, the netcdf-c library pairs up each HDF5 library API with a corresponding Codec API by invoking the relevant well-known functions
-(See Appendix E).
+(See Appendix E/a>).
This results in this table for associated codec and hdf5 libraries.
| HDF5 API | Codec API | Action
@@ -481,7 +506,7 @@ As a special case, a shared library may be created to hold
defaults for a common set of filters.
Basically, there is a specially defined function that returns
a vector of codec APIs. These defaults are used only if
-not other library provided codec information for a filter.
+no other library provides codec information for a filter.
Currently, the defaults library provides codec defaults
for Shuffle, Fletcher32, Deflate (zlib), and SZIP.
@@ -493,9 +518,9 @@ filters and to process the meta-data in Codec JSON format.
### Writing an NCZarr Container
-When writing, the user program will invoke the NetCDF API function *nc_def_var_filter*.
+When writing, the user program will invoke the NetCDF API function *nc\_def\_var\_filter*.
This function is currently defined to operate using HDF5-style id and parameters (unsigned ints).
-The netcdf-c library examines its list of known filters to find one matching the HDF5 id provided by *nc_def_var_filter*.
+The netcdf-c library examines its list of known filters to find one matching the HDF5 id provided by *nc\_def\_var\_filter*.
The set of parameters provided is stored internally.
Then during writing of data, the corresponding HDF5 filter is invoked to encode the data.
@@ -536,7 +561,7 @@ is stored in the JSON dictionary form described earlier.
The Codec style, using JSON, has the ability to provide very complex parameters that may be hard to encode as a vector of unsigned integers.
It might be desirable to consider exporting a JSON-base API out of the netcdf-c API to support user access to this complexity.
-This would mean providing some alternate version of "nc_def_var_filter" that takes a string-valued argument instead of a vector of unsigned ints.
+This would mean providing some alternate version of "nc\_def\_var\_filter" that takes a string-valued argument instead of a vector of unsigned ints.
This extension is unlikely to be implemented until a compelling use-case is encountered.
One bad side-effect of this is that we then may have two classes of plugins.
@@ -544,10 +569,54 @@ One class can be used by both HDF5 and NCZarr, and a second class that is usable
## Using The NetCDF-C Plugins
-As part of its testing, the NetCDF build process creates a number of shared libraries in the ''netcdf-c/plugins'' (or sometimes ''netcdf-c/plugins/.libs'') directory.
-If you need a filter from that set, you may be able to set ''HDF5_PLUGIN_PATH''
+As part of its testing, the NetCDF build process creates a number of shared libraries in the *netcdf-c/plugins* (or sometimes *netcdf-c/plugins/.libs*) directory.
+If you need a filter from that set, you may be able to set *HDF5\_PLUGIN\_PATH*
to point to that directory or you may be able to copy the shared libraries out of that directory to your own location.
+# Lossy One-Way Filters
+
+As of NetCDF version 4.8.2, the netcdf-c library supports
+bit-grooming filters.
+````
+Bit-grooming is a lossy compression algorithm that removes the
+bloat due to false-precision, those bits and bytes beyond the
+meaningful precision of the data. Bit Grooming is statistically
+unbiased, applies to all floating point numbers, and is easy to
+use. Bit-Grooming reduces data storage requirements by
+25-80%. Unlike its best-known competitor Linear Packing, Bit
+Grooming imposes no software overhead on users, and guarantees
+its precision throughout the whole floating point range [9].
+````
+The generic term "quantize" is used to refer collectively to the various
+bitgroom algorithms. The key thing to note about quantization is that
+it occurs at the point of writing of data only. Since its output is
+legal data, it does not need to be "de-quantized" when the data is read.
+Because of this, quantization is not part of the standard filter
+mechanism and has a separate API.
+
+The API for bit-groom is currently as follows.
+````
+int nc_def_var_quantize(int ncid, int varid, int quantize_mode, int nsd);
+int nc_inq_var_quantize(int ncid, int varid, int *quantize_modep, int *nsdp);
+````
+The *quantize_mode* argument specifies the particular algorithm.
+Currently, three are supported: NC_QUANTIZE_BITGROOM, NC_QUANTIZE_GRANULARBR,
+and NC_QUANTIZE_BITROUND. In addition quantization can be disabled using
+the value NC_NOQUANTIZE.
+
+The input to ncgen or the output from ncdump supports special attributes
+to indicate if quantization was applied to a given variable.
+These attributes have the following form.
+````
+_QuantizeBitGroomNumberOfSignificantDigits =
+or
+_QuantizeGranularBitRoundNumberOfSignificantDigits =
+or
+_QuantizeBitRoundNumberOfSignificantBits =
+````
+The value NSD is the number of significant (decimal) digits to keep.
+The value NSB is the number of significant bits to keep.
+
# Debugging {#filters_debug}
Depending on the debugger one uses, debugging plugins can be very difficult.
@@ -556,9 +625,9 @@ It may be necessary to use the old printf approach for debugging the filter itse
One case worth mentioning is when there is a dataset that is using an unknown filter.
For this situation, you need to identify what filter(s) are used in the dataset.
This can be accomplished using this command.
-````
-ncdump -s -h
-````
+
+ ncdump -s -h
+
Since ncdump is not being asked to access the data (the -h flag), it can obtain the filter information without failures.
Then it can print out the filter id and the parameters as well as the Codecs (via the -s flag).
@@ -566,34 +635,35 @@ Then it can print out the filter id and the parameters as well as the Codecs (vi
Within the netcdf-c source tree, the directory two directories contain test cases for testing dynamic filter operation.
-* __netcdf-c/nc_test4__ provides tests for testing HDF5 filters.
-* __netcdf-c/nczarr_test__ provides tests for testing NCZarr filters.
+* *netcdf-c/nc\_test4* provides tests for testing HDF5 filters.
+* *netcdf-c/nczarr\_test* provides tests for testing NCZarr filters.
-These tests are disabled if __--disable-shared__ or if __--disable-filter-tests__ is specified.
+These tests are disabled if *--disable-shared* or if *--disable-filter-tests* is specified
+or if *--disable-plugins* is specified.
## HDF5 Example {#filters_Example}
-A slightly simplified version of one of the HDF5 filter test cases is also available as an example within the netcdf-c source tree directory __netcdf-c/examples/C__.
-The test is called __filter_example.c__ and it is executed as part of the __run_examples4.sh__ shell script.
+A slightly simplified version of one of the HDF5 filter test cases is also available as an example within the netcdf-c source tree directory *netcdf-c/examples/C*.
+The test is called *filter\_example.c* and it is executed as part of the *run\_examples4.sh* shell script.
The test case demonstrates dynamic filter writing and reading.
-The files __example/C/hdf5plugins/Makefile.am__ and __example/C/hdf5plugins/CMakeLists.txt__ demonstrate how to build the hdf5 plugin for bzip2.
+The files *example/C/hdf5plugins/Makefile.am* and *example/C/hdf5plugins/CMakeLists.txt* demonstrate how to build the hdf5 plugin for bzip2.
# Notes
## Order of Invocation for Multiple Filters
-When multiple filters are defined on a variable, the order of application, when writing data to the file, is same as the order in which _nc_def_var_filter_ is called.
+When multiple filters are defined on a variable, the order of application, when writing data to the file, is same as the order in which *nc\_def\_var\_filter*is called.
When reading a file the order of application is of necessity the reverse.
There are some special cases.
1. The fletcher32 filter is always applied first, if enabled.
-2. If _nc_def_var_filter_ or _nc_def_var_deflate_ or _nc_def_var_szip_ is called multiple times with the same filter id, but possibly with different sets of parameters, then the position of that filter in the sequence of applictions does not change.
+2. If *nc\_def\_var\_filter*or *nc\_def\_var\_deflate*or *nc\_def\_var\_szip*is called multiple times with the same filter id, but possibly with different sets of parameters, then the position of that filter in the sequence of applictions does not change.
However the last set of parameters specified is used when actually writing the dataset.
3. Deflate and shuffle — these two are inextricably linked in the current API, but have quite different semantics.
- If you call _nc_def_var_deflate_ multiple times, then the previous rule applies with respect to deflate.
- However, the shuffle filter, if enabled, is ''always'' applied before applying any other filters, except fletcher32.
+ If you call *nc\_def\_var\_deflate*multiple times, then the previous rule applies with respect to deflate.
+ However, the shuffle filter, if enabled, is *always* applied before applying any other filters, except fletcher32.
4. Once a filter is defined for a variable, it cannot be removed nor can its position in the filter order be changed.
## Memory Allocation Issues
@@ -603,13 +673,12 @@ Starting with HDF5 version 1.10.*, the plugin code MUST be careful when using th
In the event that the code is allocating, reallocating, for
free'ing memory that either came from or will be exported to the
calling HDF5 library, then one MUST use the corresponding HDF5
-functions *H5allocate_memory()*, *H5resize_memory()*,
-*H5free_memory()* [5] to avoid memory failures.
+functions *H5allocate\_memory()*, *H5resize\_memory()*,
+*H5free\_memory()* [5] to avoid memory failures.
Additionally, if your filter code leaks memory, then the HDF5 library generates a failure something like this.
-````
-H5MM.c:232: H5MM_final_sanity_check: Assertion `0 == H5MM_curr_alloc_bytes_s' failed.
-````
+
+ H5MM.c:232: H5MM_final_sanity_check: Assertion `0 == H5MM_curr_alloc_bytes_s' failed.
One can look at the the code in plugins/H5Zbzip2.c and H5Zmisc.c as illustrations.
@@ -620,13 +689,22 @@ These are handled internally to (mostly) hide them so that they should not affec
Specifically, this filter may do two things.
1. Add extra parameters to the filter parameters: going from the two parameters provided by the user to four parameters for internal use.
- It turns out that the two parameters provided when calling nc_def_var_filter correspond to the first two parameters of the four parameters returned by nc_inq_var_filter.
-2. Change the values of some parameters: the value of the __options_mask__ argument is known to add additional flag bits, and the __pixels_per_block__ parameter may be modified.
+ It turns out that the two parameters provided when calling nc\_def\_var\_filter correspond to the first two parameters of the four parameters returned by nc\_inq\_var\_filter.
+2. Change the values of some parameters: the value of the *options\_mask* argument is known to add additional flag bits, and the *pixels\_per\_block* parameter may be modified.
-The reason for these changes is has to do with the fact that the szip API provided by the underlying H5Pset_szip function is actually a subset of the capabilities of the real szip implementation.
+The reason for these changes is has to do with the fact that the szip API provided by the underlying H5Pset\_szip function is actually a subset of the capabilities of the real szip implementation.
Presumably this is for historical reasons.
-In any case, if the caller uses the __nc_inq_var_szip__ or the __nc_inq_var_filter__ functions, then the parameter values returned may differ from those originally specified.
+In any case, if the caller uses the *nc\_inq\_var\_szip* or the *nc\_inq\_var\_filter* functions, then the parameter values returned may differ from those originally specified.
+
+It should also be noted that the HDF5 szip filter wrapper that
+is invoked depends on the configuration of the netcdf-c library.
+If the HDF5 installation supports szip, then the NCZarr szip
+will use the HDF5 wrapper. If HDF5 does not support szip, or HDF5
+is not enabled, then the plugins directory will contain a local
+HDF5 szip wrapper to be used by NCZarr. This can be confusing,
+but is generally transparent to the use since the plugins
+HDF5 szip wrapper was taken from the HDF5 code base.
## Supported Systems
@@ -639,9 +717,8 @@ The current matrix of OS X build systems known to work is as follows.
## Generic Plugin Build
If you do not want to use Automake or Cmake, the following has been known to work.
-````
-gcc -g -O0 -shared -o libbzip2.so -L${HDF5LIBDIR} -lhdf5_hl -lhdf5 -L${ZLIBDIR} -lz
-````
+
+ gcc -g -O0 -shared -o libbzip2.so -L${HDF5LIBDIR} -lhdf5\_hl -lhdf5 -L${ZLIBDIR} -lz
# References {#filters_References}
@@ -649,9 +726,11 @@ gcc -g -O0 -shared -o libbzip2.so -L${HDF5LIBDIR} -lhdf5_
2. https://support.hdfgroup.org/HDF5/doc/TechNotes/TechNote-HDF5-CompressionTroubleshooting.pdf
3. https://portal.hdfgroup.org/display/support/Registered+Filter+Plugins
4. https://support.hdfgroup.org/services/contributions.html#filters
-5. https://support.hdfgroup.org/HDF5/doc/RM/RM_H5.html
+5. https://support.hdfgroup.org/HDF5/doc/RM/RM\_H5.html
6. https://confluence.hdfgroup.org/display/HDF5/Filters
7. https://numcodecs.readthedocs.io/en/stable/
+8. https://github.com/ccr/ccr
+9. https://escholarship.org/uc/item/7xd1739k
# Appendix A. HDF5 Parameter Encode/Decode {#filters_appendixa}
@@ -662,7 +741,8 @@ The bzip2 compression filter, for example, expects a single integer value from z
This encodes naturally as a single unsigned integer.
Note that signed integers and single-precision (32-bit) float values also can easily be represented as 32 bit unsigned integers by proper casting to an unsigned integer so that the bit pattern is preserved.
-Simple integer values of type short or char (or the unsigned versions) can also be mapped to an unsigned integer by truncating to 16 or 8 bits respectively and then zero extending.
+Simple signed integer values of type short or char can also be mapped to an unsigned integer by truncating to 16 or 8 bits respectively and then sign extending. Similarly, unsigned 8 and 16 bit
+values can be used with zero extensions.
Machine byte order (aka endian-ness) is an issue for passing some kinds of parameters.
You might define the parameters when compressing on a little endian machine, but later do the decompression on a big endian machine.
@@ -679,9 +759,10 @@ But this will be incorrect for 64-bit values.
So, we have this situation (for HDF5 only):
-1. the 8 bytes come in as native machine order for the machine doing the call to *nc_def_var_filter*.
-2. HDF5 divides the 8 bytes into 2 four byte pieces and ensures that each piece is in network (big) endian order.
-3. When the filter is called, the two pieces are returned in the same order but with the bytes in each piece consistent with the native machine order for the machine executing the filter.
+1. the 8 bytes start as native machine order for the machine doing the call to *nc\_def\_var\_filter*.
+2. The caller divides the 8 bytes into 2 four byte pieces and passes them to *nc\_def\_var\_filter*.
+3. HDF5 takes each four byte piece and ensures that each piece is in network (big) endian order.
+4. When the filter is called, the two pieces are returned in the same order but with the bytes in each piece consistent with the native machine order for the machine executing the filter.
## Encoding Algorithms for HDF5
@@ -723,38 +804,38 @@ To support these rules, some utility programs exist and are discussed in Filter Specification Syntax.
-2. ''int ncaux_h5filterspec_parselist(const char* txt, int* formatp, size_t* nspecsp, struct NC_H5_Filterspec*** vectorp);''
+2. *int ncaux\_h5filterspec\_parselist(const char* txt, int* formatp, size\_t* nspecsp, struct NC\_H5\_Filterspec*** vectorp);*
* txt contains the text of a sequence '|' separated filter specs.
* formatp currently always returns 0.
* nspecsp will return the number of filter specifications.
* vectorp will return a pointer to a vector of pointers to filter specification instances — the caller must free.
This function parses a sequence of filter specifications each separated by a '|' character.
-The text between '|' separators must be parsable by __ncaux_h5filterspec_parse__.
-3. ''void ncaux_h5filterspec_free(struct NC_H5_Filterspec* f);''
- * f is a pointer to an instance of ````struct NC_H5_Filterspec````
+The text between '|' separators must be parsable by *ncaux\_h5filterspec\_parse*.
+3. *void ncaux\_h5filterspec\_free(struct NC\_H5\_Filterspec* f);*
+ * f is a pointer to an instance of *struct NC\_H5\_Filterspec*
Typically this was returned as an element of the vector returned
- by __ncaux_h5filterspec_parselist__.
+ by *\_ncaux\_h5filterspec\_parselist*.
This reclaims the parameters of the filter spec object as well as the object itself.
-4. ''int ncaux_h5filterspec_fix8(unsigned char* mem8, int decode);''
+4. *int ncaux\_h5filterspec\_fix8(unsigned char* mem8, int decode);*
* mem8 is a pointer to the 8-byte value either to fix.
* decode is 1 if the function should apply the 8-byte decoding algorithm
else apply the encoding algorithm.
This function implements the 8-byte conversion algorithms for HDF5.
-Before calling *nc_def_var_filter* (unless *NC_parsefilterspec* was used), the client must call this function with the decode argument set to 0.
+Before calling *nc\_def\_var\_filter* (unless *NC\_parsefilterspec* was used), the client must call this function with the decode argument set to 0.
Inside the filter code, this function should be called with the decode argument set to 1.
-Examples of the use of these functions can be seen in the test program ''nc_test4/tst_filterparser.c''.
+Examples of the use of these functions can be seen in the test program *nc\_test4/tst\_filterparser.c*.
-Some of the above functions use a C struct defined in _netcdf_filter.h_.
+Some of the above functions use a C struct defined in *netcdf\_filter.h\_.
The definition of that struct is as follows.
````
typedef struct NC_H5_Filterspec {
@@ -767,19 +848,18 @@ This struct in effect encapsulates all of the information about and HDF5 formatt
# Appendix C. Build Flags for Detecting the Filter Mechanism {#filters_appendixc}
-The include file _netcdf_meta.h contains the following definition.
+The include file *netcdf\_meta.h* contains the following definition.
````
-#define NC_HAS_MULTIFILTERS 1
+ #define NC_HAS_MULTIFILTERS 1
````
+This, in conjunction with the error code *NC\_ENOFILTER* in *netcdf.h* can be used to see what filter mechanism is in place as described in the section on incompatibities.
-This, in conjunction with the error code _NC_ENOFILTER_ in _netcdf.h_ can be used to see what filter mechanism is in place as described in the section on incompatibities.
-
-1. !defined(NC_ENOFILTER) && !defined(NC_HAS_MULTIFILTERS) — indicates that the old pre-4.7.4 mechanism is in place.
+1. !defined(NC\_ENOFILTER) && !defined(NC\_HAS\_MULTIFILTERS) — indicates that the old pre-4.7.4 mechanism is in place.
It does not support multiple filters.
-2. defined(NC_ENOFILTER) && !defined(NC_HAS_MULTIFILTERS) — indicates that the 4.7.4 mechanism is in place.
- It does support multiple filters, but the error return codes for _nc_inq_var_filter_ are different and the filter spec parser functions are in a different location with different names.
-3. defined(NC_ENOFILTER) && defined(NC_HAS_MULTIFILTERS) — indicates that the multiple filters are supported, and that _nc_inq_var_filter_ returns a filterid of zero to indicate that a variable has no filters.
- Also, the filter spec parsers have the names and signatures described in this document and are define in _netcdf_aux.h_.
+2. defined(NC\_ENOFILTER) && !defined(NC\_HAS\_MULTIFILTERS) — indicates that the 4.7.4 mechanism is in place.
+ It does support multiple filters, but the error return codes for *nc\_inq\_var\_filter* are different and the filter spec parser functions are in a different location with different names.
+3. defined(NC\_ENOFILTER) && defined(NC\_HAS\_MULTIFILTERS) — indicates that the multiple filters are supported, and that *nc\_inq\_var\_filter* returns a filterid of zero to indicate that a variable has no filters.
+ Also, the filter spec parsers have the names and signatures described in this document and are define in *netcdf\_aux.h*.
# Appendix D. BNF for Specifying Filters in Utilities {#filters_appendixd}
@@ -800,24 +880,25 @@ parameter: unsigned32
where
unsigned32: <32 bit unsigned integer>
````
+
# Appendix E. Codec API {#filters_appendixe}
The Codec API mirrors the HDF5 API closely. It has one well-known function that can be invoked to obtain information about the Codec as well as pointers to special functions to perform conversions.
## The Codec Plugin API
-### NCZ_get_codec_info
+### NCZ\_get\_codec\_info
This function returns a pointer to a C struct that provides detailed information about the codec plugin.
#### Signature
````
-void* NCZ_get_codec_info(void);
+ void* NCZ_get_codec_info(void);
````
-The value returned is actually of type ''struct NCZ_codec_t'',
-but is of type ''void*'' to allow for extensions.
+The value returned is actually of type *struct NCZ\_codec\_t*,
+but is of type *void\** to allow for extensions.
-### NCZ_codec_t
+### NCZ\_codec\_t
````
typedef struct NCZ_codec_t {
int version; /* Version number of the struct */
@@ -835,143 +916,192 @@ typedef struct NCZ_codec_t {
The semantics of the non-function fields is as follows:
-1. ''version'' -- Version number of the struct.
-2. ''sort'' -- Format of remainder of the struct; currently always NCZ_CODEC_HDF5.
-3. ''codecid'' -- The name/id of the codec.
-4. ''hdf5id'' -- The corresponding hdf5 id.
+1. *version* — Version number of the struct.
+2. *sort* — Format of remainder of the struct; currently always NCZ\_CODEC\_HDF5.
+3. *codecid* — The name/id of the codec.
+4. *hdf5id* — The corresponding hdf5 id.
-### NCZ_codec_to_hdf5
+### NCZ\_codec\_to\_hdf5
Given a JSON Codec representation, it will return a corresponding vector of unsigned integers representing the
visible parameters.
#### Signature
- int (*NCZ_codec_to_hdf)(const char* codec, int* nparamsp, unsigned** paramsp);
-
+````
+ int NCZ_codec_to_hdf(const char* codec, int* nparamsp, unsigned** paramsp);
+````
#### Arguments
-1. codec -- (in) ptr to JSON string representing the codec.
-2. nparamsp -- (out) store the length of the converted HDF5 unsigned vector
-3. paramsp -- (out) store a pointer to the converted HDF5 unsigned vector; caller must free the returned vector. Note the double indirection.
+1. codec — (in) ptr to JSON string representing the codec.
+2. nparamsp — (out) store the length of the converted HDF5 unsigned vector
+3. paramsp — (out) store a pointer to the converted HDF5 unsigned vector; caller must free the returned vector. Note the double indirection.
Return Value: a netcdf-c error code.
-### NCZ_hdf5_to_codec
+### NCZ\_hdf5\_to\_codec
Given an HDF5 visible parameters vector of unsigned integers and its length,
return a corresponding JSON codec representation of those visible parameters.
#### Signature
- int (*NCZ_hdf5_to_codec)(int ncid, int varid, size_t nparams, const unsigned* params, char** codecp);
-
+````
+ int NCZ_hdf5_to_codec)(int ncid, int varid, size_t nparams, const unsigned* params, char** codecp);
+````
#### Arguments
-1. ncid -- the variables' containing group
-2. varid -- the containing variable
-3. nparams -- (in) the length of the HDF5 visible parameters vector
-4. params -- (in) pointer to the HDF5 visible parameters vector.
-5. codecp -- (out) store the string representation of the codec; caller must free.
+1. ncid — the variables' containing group
+2. varid — the containing variable
+3. nparams — (in) the length of the HDF5 visible parameters vector
+4. params — (in) pointer to the HDF5 visible parameters vector.
+5. codecp — (out) store the string representation of the codec; caller must free.
Return Value: a netcdf-c error code.
-### NCZ_modify_parameters
+### NCZ\_modify\_parameters
Extract environment information from the (ncid,varid) and use it to convert a set of visible parameters
to a set of working parameters; also provide option to modify visible parameters.
#### Signature
- int (*NCZ_modify_parameters)(int ncid, int varid, size_t* vnparamsp, unsigned** vparamsp, size_t* wnparamsp, unsigned** wparamsp);
-
+````
+ int NCZ_modify_parameters(int ncid, int varid, size_t* vnparamsp, unsigned** vparamsp, size_t* wnparamsp, unsigned** wparamsp);
+````
#### Arguments
-1. ncid -- (in) group id containing the variable.
-2. varid -- (in) the id of the variable to which this filter is being attached.
-3. vnparamsp -- (in/out) the count of visible parameters
-4. vparamsp -- (in/out) the set of visible parameters
-5. wnparamsp -- (out) the count of working parameters
-4. wparamsp -- (out) the set of working parameters
+1. ncid — (in) group id containing the variable.
+2. varid — (in) the id of the variable to which this filter is being attached.
+3. vnparamsp — (in/out) the count of visible parameters
+4. vparamsp — (in/out) the set of visible parameters
+5. wnparamsp — (out) the count of working parameters
+4. wparamsp — (out) the set of working parameters
Return Value: a netcdf-c error code.
-### NCZ_codec_initialize
+### NCZ\_codec\_initialize
Some compressors may require library initialization.
This function is called as soon as a shared library is loaded and matched with an HDF5 filter.
#### Signature
- int (*NCZ_codec_initialize)(void);
-
+````
+ int NCZ_codec_initialize)(void);
+````
Return Value: a netcdf-c error code.
-### NCZ_codec_finalize
+### NCZ\_codec\_finalize
Some compressors (like blosc) require invoking a finalize function in order to avoid memory loss.
-This function is called during a call to ''nc_finalize'' to do any finalization.
-If the client code does not invoke ''nc_finalize'' then memory checkers may complain about lost memory.
+This function is called during a call to *nc\_finalize* to do any finalization.
+If the client code does not invoke *nc\_finalize* then memory checkers may complain about lost memory.
#### Signature
- int (*NCZ_codec_finalize)(void);
-
+````
+ int NCZ_codec_finalize)(void);
+````
Return Value: a netcdf-c error code.
## Multi-Codec API
-As an aid to clients, it is convenient if a single shared library can provide multiple ''NCZ_code_t'' instances at one time.
+As an aid to clients, it is convenient if a single shared library can provide multiple *NCZ\_code\_t* instances at one time.
This API is not intended to be used by plugin developers.
A shared library must only export this function.
-### NCZ_codec_info_defaults
+### NCZ\_codec\_info\_defaults
-Return a NULL terminated vector of pointers to instances of ''NCZ_codec_t''.
+Return a NULL terminated vector of pointers to instances of *NCZ\_codec\_t*.
#### Signature
+````
void* NCZ_codec_info_defaults(void);
-
-The value returned is actually of type ''NCZ_codec_t**'',
-but is of type ''void*'' to allow for extensions.
+````
+The value returned is actually of type *NCZ\_codec\_t***,
+but is of type *void** to allow for extensions.
The list of returned items are used to try to provide defaults
for any HDF5 filters that have no corresponding Codec.
This is for internal use only.
-# Appendix F. Pre-built Filters
-
-As part of the overall build process, a number of filters are built as shared libraries in the "plugins" directory
-— in that directory or the "plugins/.libs" subdirectory.
+# Appendix F. Standard Filters
-An option exists to allow some of those filters to be installed into a user-specified directory. The relevant options are as follows:
+Support for a select set of standard filters is built into the NetCDF API.
+Generally, they are accessed using the following generic API, where XXXX is
+the filter name. As a rule, the names are those used in the HDF5 filter ID naming authority [4] or the NumCodecs naming authority [7].
````
-./configure: --with-plugin-dir=
-cmake: -DPLUGIN_INSTALL_DIR=
+int nc_def_var_XXXX(int ncid, int varid, unsigned filterid, size_t nparams, unsigned* params);
+int nc_inq_var_XXXX(int ncid, int varid, int* hasfilter, size_t* nparamsp, unsigned* params);
````
-If the value of the environment variable "HDF5_PLUGIN_PATH" is a single directory, then
-a good value for the install directory is "$HDF5_PLUGIN_PATH", so for example:
-````
-./configure ... --with-plugin-dir="$HDF5_PLUGIN_DIR"
-````
-
-If this option is specified, then as part of the "install" build action,
-a specified set of filter shared libraries will be copied into the specified directory.
-Any existing library of the same name will be overwritten. If the specified directory
-itself does not exist, then it will be created.
-
-Currently, if the following filters are available, they will be installed;
-* ''libh5bzip2.so'' -- an HDF5 filter wrapper for bzip2 compression
-* ''libh5blosc.so'' -- an HDF5 filter wrapper for blosc compression
-* ''libh5zstd.so'' -- an HDF5 filter wrapper for zstandardcompression
-
-If the user is using NCZarr filters, then the following additional filters will be installed.
-* libh5shuffle.so -- shuffle filter
-* libh5fletcher32.so -- fletcher32 checksum
-* libh5deflate.so -- deflate compression
-* libh5szip.so -- szip compression
-* libnczdefaults.so -- provide NCZarr support for shuffle, fletcher32, and deflate.
-* libnczszip.so -- provide NCZarr support for szip.
+The first function inserts the specified filter into the filter chain for a given variable.
+The second function queries the given variable to see if the specified function
+is in the filter chain for that variable. The *hasfilter* argument is set
+to one if the filter is in the chain and zero otherwise.
+As is usual with the netcdf API, one is expected to call this function twice.
+The first time to set *nparamsp* and the second to get the parameters in the client-allocated memory argument *params*.
+Any of these arguments can be NULL, in which case no value is returned.
+
+Note that NetCDF inherits four filters from HDF5, namely shuffle, fletcher32, deflate (zlib), and szip. The API's for these do not conform to the above API.
+So aside from those four, the current set of standard filters is as follows.
+
+| Filter Name | Filter ID | Reference
+ |
|---|
| zstandard | 32015 | https://facebook.github.io/zstd/
+ | | bzip2 | 307 | https://sourceware.org/bzip2/
+ |
-The shuffle, fletcher32, deflate, and szip filters in this case will be ignored by HDF5 and only used by the NCZarr code.
-Note that if you disable HDF5 support, but leave NCZarr support enabled, then all of the above filters
-should continue to work.
+# Appendix G. Finding Filters
+
+A major problem for filter users is finding an implementation for a filter.
+There are several ways to do this.
+
+* **HDF5 Assigned Filter Identifiers Repository [3]** —
+HDF5 maintains a page of standard filter identifiers along with
+additional contact information. This often includes a pointer
+to source code.
+
+* **Community Codec Repository** —
+The Community Codec Repository (CCR) project [8] provides
+filters, including HDF5 wrappers, for a number of filters.
+You can install this library to get access to these supported filters.
+It does not currently include the required NCZarr Codec API,
+so they are only usable with netcdf-4. This will change in the future.
+
+* **NetCDF-C Test Plugins Directory** —
+As part of the overall build process, a number of filters are built as shared libraries in the "plugins" directory.
+They may be in that directory or the "plugins/.libs" subdirectory.
+It may be possible for users to utilize some of those libraries to provide filter support for general use.
+
+
+ If the user is using NCZarr filters, then the plugins directory has at least the following shared libraries
+ * libh5shuffle.so — shuffle filter
+ * libh5fletcher32.so — fletcher32 checksum
+ * libh5deflate.so — deflate compression
+ * libnczdefaults.so — provide NCZarr support for shuffle, fletcher32, and deflate.
+ * *libh5bzip2.so* — an HDF5 filter for bzip2 compression
+ * *libh5blosc.so* — an HDF5 filter for blosc compression
+ * *libh5zstd.so* — an HDF5 filter for zstandard compression
+
+ The shuffle, fletcher32, and deflate filters in this case will
+ be ignored by HDF5 and only used by the NCZarr code. But in
+ order to use them, it needs additional Codec capabilities
+ provided by the libnczdefauts.so shared library. Note also that
+ if you disable HDF5 support, but leave NCZarr support enabled,
+ then all of the above filters should continue to work.
+
+## HDF5_PLUGIN_PATH
+
+At the moment, NetCDF uses the existing HDF5 environment variable
+*HDF5\_PLUGIN\_PATH* to locate the directories in which filter wrapper
+shared libraries are located. This is used both for the HDF5 filter
+wrappers but also the NCZarr codec wrappers.
+
+*HDF5\_PLUGIN\_PATH* is a typical Windows or Unix style
+path-list. That is it is a sequence of absolute directory paths
+separated by a specific separator character. For Windows, the
+separator character is a semicolon (';') and for Unix, it is a a
+colon (':'). For convenience, NCZarr will also accept the
+semicolon separator for Unix.
+
+So, the user can add the CCR and/or the plugins directory to
+the *HDF5\_PLUGIN\_PATH* environment variable to allow the netcdf-c
+library to locate wrappers.
# Point of Contact {#filters_poc}
-__Author__: Dennis Heimbigner
-__Email__: dmh at ucar dot edu
-__Initial Version__: 1/10/2018
-__Last Revised__: 7/17/2021
-
+*Author*: Dennis Heimbigner
+*Email*: dmh at ucar dot edu
+*Initial Version*: 1/10/2018
+*Last Revised*: 3/14/2022
diff --git a/docs/internal.md b/docs/internal.md
new file mode 100644
index 0000000000..05c35257c6
--- /dev/null
+++ b/docs/internal.md
@@ -0,0 +1,639 @@
+Notes On the Internals of the NetCDF-C Library
+============================
+
+
+# Notes On the Internals of the NetCDF-C Library {#intern_head}
+
+\tableofcontents
+
+This document attempts to record important information about
+the internal architecture and operation of the netcdf-c library.
+
+# 1. Including C++ Code in the netcdf-c Library {#intern_c++}
+
+The state of C compiler technology has reached the point where
+it is possible to include C++ code into the netcdf-c library
+code base. Two examples are:
+
+1. The AWS S3 SDK wrapper *libdispatch/ncs3sdk.cpp* file.
+2. The TinyXML wrapper *ncxml\_tinyxml2.cpp* file.
+
+However there are some consequences that must be handled for this to work.
+Specifically, the compiler must be told that the C++ runtime is needed
+in the following ways.
+
+## Modifications to *lib\_flags.am*
+Suppose we have a flag *ENABLE\_XXX* where that XXX
+feature entails using C++ code. Then the following must be added
+to *lib\_flags.am*
+````
+if ENABLE_XXX
+AM_LDFLAGS += -lstdc++
+endif
+````
+
+## Modifications to *libxxx/Makefile.am*
+
+The Makefile in which the C++ code is included and compiled
+(assumed here to be the *libxxx* directory) must have this set.
+````
+AM_CXXFLAGS = -std=c++11
+````
+It is possible that other values (e.g. *-std=c++14*) may also work.
+
+# 2. Managing instances of complex data types
+
+For a long time, there have been known problems with the
+management of complex types containing VLENs. This also
+involves the string type because it is stored as a VLEN of
+chars.
+
+The term complex type refers to any type that directly or
+recursively references a VLEN type. So an array of VLENS, a
+compound with a VLEN field, and so on.
+
+In order to properly handle instances of these complex types, it
+is necessary to have function that can recursively walk
+instances of such types to perform various actions on them. The
+term "deep" is also used to mean recursive.
+
+Two deep walking operations are provided by the netcdf-c library
+to aid in managing instances of complex structures.
+* free'ing an instance of the complex type
+* copying an instance of the complex type.
+
+Previously The netcdf-c library only did shallow free and shallow copy of
+complex types. This meant that only the top level was properly
+free'd or copied, but deep internal blocks in the instance were
+not touched. This led to a host of memory leaks and failures
+when the deep data was effectively shared between the netcdf-c library
+internally and the user's data.
+
+Note that the term "vector" is used to mean a contiguous (in
+memory) sequence of instances of some type. Given an array with,
+say, dimensions 2 X 3 X 4, this will be stored in memory as a
+vector of length 2*3*4=24 instances.
+
+The use cases are primarily these.
+
+## nc\_get\_vars
+Suppose one is reading a vector of instances using nc\_get\_vars
+(or nc\_get\_vara or nc\_get\_var, etc.). These functions will
+return the vector in the top-level memory provided. All
+interior blocks (form nested VLEN or strings) will have been
+dynamically allocated. Note that computing the size of the vector
+may be tricky because the strides must be taken into account.
+
+After using this vector of instances, it is necessary to free
+(aka reclaim) the dynamically allocated memory, otherwise a
+memory leak occurs. So, the recursive reclaim function is used
+to walk the returned instance vector and do a deep reclaim of
+the data.
+
+Currently functions are defined in netcdf.h that are supposed to
+handle this: nc\_free\_vlen(), nc\_free\_vlens(), and
+nc\_free\_string(). Unfortunately, these functions only do a
+shallow free, so deeply nested instances are not properly
+handled by them. They are marked in the description as
+deprecated in favor of the newer recursive function.
+
+## nc\_put\_vars
+
+Suppose one is writing a vector of instances using nc\_put\_vars
+(or nc\_put\_vara or nc\_put\_var, etc.). These functions will
+write the contents of the vector to the specified variable.
+Note that internally, the data passed to the nc\_put\_xxx function is
+immediately written so there is no need to copy it internally. But the
+caller may need to reclaim the vector of data that was created and passed
+in to the nc\_put\_xxx function.
+
+After writing this vector of instances, and assuming it was dynamically
+created, at some point it will be necessary to reclaim that data.
+So again, the recursive reclaim function can be used
+to walk the returned instance vector and do a deep reclaim of
+the data.
+
+## nc\_put\_att
+Suppose one is writing a vector of instances as the data of an attribute
+using, say, nc\_put\_att.
+
+Internally, the incoming attribute data must be copied and stored
+so that changes/reclamation of the input data will not affect
+the attribute. Note that this copying behavior is different from
+writing to a variable, where the data is written immediately.
+
+Again, the code inside the netcdf library used to use only shallow copying
+rather than deep copy. As a result, one saw effects such as described
+in Github Issue https://github.com/Unidata/netcdf-c/issues/2143.
+
+Also, after defining the attribute, it may be necessary for the user
+to free the data that was provided as input to nc\_put\_att() as in the
+nc\_put\_xxx functions (previously described).
+
+## nc\_get\_att
+Suppose one is reading a vector of instances as the data of an attribute
+using, say, nc\_get\_att.
+
+Internally, the existing attribute data must be copied and returned
+to the caller, and the caller is responsible for reclaiming
+the returned data.
+
+Again, the code inside the netcdf library used to only do shallow copying
+rather than deep copy. So this could lead to memory leaks and errors
+because the deep data was shared between the library and the user.
+
+## New Instance Walking API
+
+Proper recursive functions were added to the netcdf-c library to
+provide reclaim and copy functions and use those as needed.
+These functions are defined in libdispatch/dinstance.c and their
+signatures are defined in include/netcdf.h. For back
+compatibility, corresponding "ncaux\_XXX" functions are defined
+in include/netcdf\_aux.h.
+````
+int nc_reclaim_data(int ncid, nc_type xtypeid, void* memory, size_t count);
+int nc_reclaim_data_all(int ncid, nc_type xtypeid, void* memory, size_t count);
+int nc_copy_data(int ncid, nc_type xtypeid, const void* memory, size_t count, void* copy);
+int nc_copy_data_all(int ncid, nc_type xtypeid, const void* memory, size_t count, void** copyp);
+````
+There are two variants. The first two, nc\_reclaim\_data() and
+nc\_copy\_data(), assume the top-level vector is managed by the
+caller. For reclaim, this is so the user can use, for example, a
+statically allocated vector. For copy, it assumes the user
+provides the space into which the copy is stored.
+
+The second two, nc\_reclaim\_data\_all() and
+nc\_copy\_data\_all(), allows the functions to manage the
+top-level. So for nc\_reclaim\_data\_all, the top level is
+assumed to be dynamically allocated and will be free'd by
+nc\_reclaim\_data\_all(). The nc\_copy\_data\_all() function
+will allocate the top level and return a pointer to it to the
+user. The user can later pass that pointer to
+nc\_reclaim\_data\_all() to reclaim the instance(s).
+
+# Internal Changes
+The netcdf-c library internals are changed to use the proper reclaim
+and copy functions. This also allows some simplification of the code
+since the stdata and vldata fields of NC\_ATT\_INFO are no longer needed.
+Currently this is commented out using the SEPDATA \#define macro.
+When the bugs are found and fixed, all this code will be removed.
+
+## Optimizations
+
+In order to make these functions as efficient as possible, it is
+desirable to classify all types as to whether or not they contain
+variable-size data. If a type is fixed sized (i.e. does not contain
+variable-size data) then it can be freed or copied as a single chunk.
+This significantly increases the performance for such types.
+For variable-size types, it is necessary to walk each instance of the type
+and recursively reclaim or copy it. As another optimization,
+if the type is a vector of strings, then the per-instance walk can be
+sped up by doing the reclaim or copy inline.
+
+The rules for classifying types as fixed or variable size are as follows.
+
+1. All atomic types, except string, are fixed size.
+2. All enum type and opaque types are fixed size.
+3. All string types and VLEN types are variable size.
+4. A compound type is fixed size if all of the types of its
+ fields are fixed size. Otherwise it has variable size.
+
+The classification of types can be made at the time the type is defined
+or is read in from some existing file. The reclaim and copy functions
+use this information to speed up the handling of fixed size types.
+
+# Warnings
+
+1. The new API functions require that the type information be
+ accessible. This means that you cannot use these functions
+ after the file has been closed. After the file is closed, you
+ are on your own.
+
+2. There is still one known failure that has not been solved; it is
+ possibly an HDF5 memory leak. All the failures revolve around
+ some variant of this .cdl file. The proximate cause of failure is
+ the use of a VLEN FillValue.
+````
+ netcdf x {
+ types:
+ float(*) row_of_floats ;
+ dimensions:
+ m = 5 ;
+ variables:
+ row_of_floats ragged_array(m) ;
+ row_of_floats ragged_array:_FillValue = {-999} ;
+ data:
+ ragged_array = {10, 11, 12, 13, 14}, {20, 21, 22, 23}, {30, 31, 32},
+ {40, 41}, _ ;
+ }
+````
+
+# 3. Inferring File Types
+
+As described in the companion document -- docs/dispatch.md --
+when nc\_create() or nc\_open() is called, it must figure out what
+kind of file is being created or opened. Once it has figured out
+the file kind, the appropriate "dispatch table" can be used
+to process that file.
+
+## The Role of URLs
+
+Figuring out the kind of file is referred to as model inference
+and is, unfortunately, a complicated process. The complication
+is mostly a result of allowing a path argument to be a URL.
+Inferring the file kind from a URL requires deep processing of
+the URL structure: the protocol, the host, the path, and the fragment
+parts in particular. The query part is currently not used because
+it usually contains information to be processed by the server
+receiving the URL.
+
+The "fragment" part of the URL may be unfamiliar.
+The last part of a URL may optionally contain a fragment, which
+is syntactically of this form in this pseudo URL specification.
+````
+:///?#
+````
+The form of the fragment is similar to a query and takes this general form.
+````
+'#'=&=&...
+````
+The key is a simple name, the value is any sequence of characters,
+although URL special characters such as '&' must be URL encoded in
+the '%XX' form where each X is a hexadecimal digit.
+An example might look like this non-sensical example:
+````
+https://host.com/path#mode=nczarr,s3&bytes
+````
+It is important to note that the fragment part is not intended to be
+passed to the server, but rather is processed by the client program.
+It is this property that allows the netcdf-c library to use it to
+pass information deep into the dispatch table code that is processing the
+URL.
+
+## Model Inference Inputs
+
+The inference algorithm is given the following information
+from which it must determine the kind of file being accessed.
+
+### Mode
+
+The mode is a set of flags that are passed as the second
+argument to nc\_create and nc\_open. The set of flags is define in
+the netcdf.h header file. Generally it specifies the general
+format of the file: netcdf-3 (classic) or netcdf-4 (enhanced).
+Variants of these can also be specified, e.g. 64-bit netcdf-3 or
+classic netcdf-4.
+In the case where the path argument is a simple file path,
+using a mode flag is the most common mechanism for specifying
+the model.
+
+### Path
+The file path, the first argument to nc\_create and nc\_open,
+Can be either a simple file path or a URL.
+If it is a URL, then it will be deeply inspected to determine
+the model.
+
+### File Contents
+When the contents of a real file are available,
+the contents of the file can be used to determine the dispatch table.
+As a rule, this is likely to be useful only for *nc\_open*.
+It also requires access to functions that can open and read at least
+the initial part of the file.
+As a rule, the initial small prefix of the file is read
+and examined to see if it matches any of the so-called
+"magic numbers" that indicate the kind of file being read.
+
+### Open vs Create
+Is the file being opened or is it being created?
+
+### Parallelism
+Is parallel IO available?
+
+## Model Inference Outputs
+The inferencing algorithm outputs two pieces of information.
+
+1. model - this is used by nc\_open and nc\_create to choose the dispatch table.
+2. newpath - in some case, usually URLS, the path may be rewritten to include extra information for use by the dispatch functions.
+
+The model output is actually a struct containing two fields:
+
+1. implementation - this is a value from the NC\_FORMATX\_xxx
+ values in netcdf.h. It generally determines the dispatch
+ table to use.
+2. format -- this is an NC\_FORMAT\_xxx value defining, in effect,
+ the netcdf-format to which the underlying format is to be
+ translated. Thus it can tell the netcdf-3 dispatcher that it
+ should actually implement CDF5 rather than standard netcdf classic.
+
+## The Inference Algorithm
+
+The construction of the model is primarily carried out by the function
+*NC\_infermodel()* (in *libdispatch/dinfermodel.c).
+It is given the following parameters:
+1. path -- (IN) absolute file path or URL
+2. modep -- (IN/OUT) the set of mode flags given to *NC\_open* or *NC\_create*.
+3. iscreate -- (IN) distinguish open from create.
+4. useparallel -- (IN) indicate if parallel IO can be used.
+5. params -- (IN/OUT) arbitrary data dependent on the mode and path.
+6. model -- (IN/OUT) place to store inferred model.
+7. newpathp -- (OUT) the canonical rewrite of the path argument.
+
+As a rule, these values are used in the this order of preference
+to infer the model.
+
+1. file contents -- highest precedence
+2. url (if it is one) -- using the "mode=" key in the fragment (see below).
+3. mode flags
+4. default format -- lowest precedence
+
+The sequence of steps is as follows.
+
+### URL processing -- processuri()
+
+If the path appears to be a URL, then it is parsed
+and processed by the processuri function as follows.
+
+1. Protocol --
+The protocol is extracted and tested against the list of
+legal protocols. If not found, then it is an error.
+If found, then it is replaced by a substitute -- if specified.
+So, for example, the protocol "dods" is replaced the protocol "http"
+(note that at some point "http" will be replaced with "https").
+Additionally, one or more "key=value" strings is appended
+to the existing fragment of the url. So, again for "dods",
+the fragment is extended by the string "mode=dap2".
+Thus replacing "dods" does not lose information, but rather transfers
+it to the fragment for later use.
+
+2. Fragment --
+After the protocol is processed, the initial fragment processing occurs
+by converting it to a list data structure of the form
+````
+ {,,,,,....}
+````
+
+### Macro Processing -- processmacros()
+
+If the fragment list produced by processuri() is non-empty, then
+it is processed for "macros". Notice that if the original path
+was not a URL, then the fragment list is empty and this
+processing will be bypassed. In any case, It is convenient to
+allow some singleton fragment keys to be expanded into larger
+fragment components. In effect, those singletons act as
+macros. They can help to simplify the user's URL. The term
+singleton means a fragment key with no associated value:
+"#bytes", for example.
+
+The list of fragments is searched looking for keys whose
+value part is NULL or the empty string. Then the table
+of macros is searched for that key and if found, then
+a key and values is appended to the fragment list and the singleton
+is removed.
+
+### Mode Inference -- processinferences()
+
+This function just processes the list of values associated
+with the "mode" key. It is similar to a macro in that
+certain mode values are added or removed based on tables
+of "inferences" and "negations".
+Again, the purpose is to allow users to provide simplified URL fragments.
+
+The list of mode values is repeatedly searched and whenever a value
+is found that is in the "modeinferences" table, then the associated inference value
+is appended to the list of mode values. This process stops when no changes
+occur. This form of inference allows the user to specify "mode=zarr"
+and have it converted to "mode=nczarr,zarr". This avoids the need for the
+dispatch table code to do the same inference.
+
+After the inferences are made, The list of mode values is again
+repeatedly searched and whenever a value
+is found that is in the "modenegations" table, then the associated negation value
+is removed from the list of mode values, assuming it is there. This process stops when no changes
+occur. This form of inference allows the user to make sure that "mode=bytes,nczarr"
+has the bytes mode take precedence by removing the "nczarr" value. Such illegal
+combinations can occur because of previous processing steps.
+
+### Fragment List Normalization
+As the fragment list is processed, duplicates appear with the same key.
+A function -- cleanfragments() -- is applied to clean up the fragment list
+by coalesing the values of duplicate keys and removing duplicate key values.
+
+### S3 Rebuild
+If the URL is determined to be a reference to a resource on the Amazon S3 cloud,
+then the URL needs to be converted to what is called "path format".
+There are four S3 URL formats:
+
+1. Virtual -- ````https://.s3..amazonaws.com/````
+2. Path -- ````https://s3..amazonaws.com//````
+3. S3 -- ````s3:///````
+4. Other -- ````https:////````
+
+The S3 processing converts all of these to the Path format. In the "S3" format case
+it is necessary to find or default the region from examining the ".aws" directory files.
+
+### File Rebuild
+If the URL protocol is "file" and its path is a relative file path,
+then it is made absolute by prepending the path of the current working directory.
+
+In any case, after S3 or File rebuilds, the URL is completely
+rebuilt using any modified protocol, host, path, and
+fragments. The query is left unchanged in the current algorithm.
+The resulting rebuilt URL is passed back to the caller.
+
+### Mode Key Processing
+The set of values of the fragment's "mode" key are processed one by one
+to see if it is possible to determine the model.
+There is a table for format interpretations that maps a mode value
+to the model's implementation and format. So for example,
+if the mode value "dap2" is encountered, then the model
+implementation is set to NC\_FORMATX\_DAP2 and the format
+is set to NC\_FORMAT\_CLASSIC.
+
+### Non-Mode Key Processing
+If processing the mode does not tell us the implementation, then
+all other fragment keys are processed to see if the implementaton
+(and format) can be deduced. Currently this does nothing.
+
+### URL Defaults
+If the model is still not determined and the path is a URL, then
+the implementation is defaulted to DAP2. This is for back
+compatibility when all URLS implied DAP2.
+
+### Mode Flags
+In the event that the path is not a URL, then it is necessary
+to use the mode flags and the isparallel arguments to choose a model.
+This is just a straight forward flag checking exercise.
+
+### Content Inference -- check\_file\_type()
+If the path is being opened (as opposed to created), then
+it may be possible to actually read the first few bytes of the
+resource specified by the path and use that to determine the
+model. If this succeeds, then it takes precedence over
+all other model inferences.
+
+### Flag Consistency
+Once the model is known, then the set of mode flags
+is modified to be consistent with that information.
+So for example, if DAP2 is the model, then all netcdf-4 mode flags
+and some netcdf-3 flags are removed from the set of mode flags
+because DAP2 provides only a standard netcdf-classic format.
+
+# 4. Adding a Standard Filter
+
+The standard filter system extends the netcdf-c library API to
+support a fixed set of "standard" filters. This is similar to the
+way that deflate and szip are currently supported.
+For background, the file filter.md should be consulted.
+
+In general, the API for a standard filter has the following prototypes.
+The case of zstandard (libzstd) is used as an example.
+````
+int nc_def_var_zstandard(int ncid, int varid, int level);
+int nc_inq_var_zstandard(int ncid, int varid, int* has_filterp, int* levelp);
+````
+So generally the API has the ncid and the varid as fixed, and then
+a list of parameters specific to the filter -- level in this case.
+For the inquire function, there is an additional argument -- has_filterp --
+that is set to 1 if the filter is defined for the given variable
+and is 0 if not.
+The remainder of the inquiry parameters are pointers to memory
+into which the parameters are stored -- levelp in this case.
+
+It is important to note that including a standard filter still
+requires three supporting objects:
+
+1. The implementing library for the filter. For example,
+ libzstd must be installed in order to use the zstandard
+ API.
+2. A HDF5 wrapper for the filter must be installed in the
+ directory pointed to by the HDF5_PLUGIN_PATH environment
+ variable.
+3. (Optional) An NCZarr Codec implementation must be installed
+ in the the HDF5_PLUGIN_PATH directory.
+
+## Adding a New Standard Filter
+
+The implementation of a standard filter must be loaded from one
+of several locations.
+
+1. It can be part of libnetcdf.so (preferred),
+2. it can be loaded as part of the client code,
+3. or it can be loaded as part of an external library such as libccr.
+
+However, the three objects listed above need to be
+stored in the HDF5_PLUGIN_DIR directory, so adding a standard
+filter still requires modification to the netcdf build system.
+This limitation may be lifted in the future.
+
+### Build Changes
+In order to detect a standard library, the following changes
+must be made for Automake (configure.ac/Makefile.am)
+and CMake (CMakeLists.txt)
+
+#### Configure.ac
+Configure.ac must have a block that similar to this that locates
+the implementing library.
+````
+# See if we have libzstd
+AC_CHECK_LIB([zstd],[ZSTD_compress],[have_zstd=yes],[have_zstd=no])
+if test "x$have_zstd" = "xyes" ; then
+ AC_SEARCH_LIBS([ZSTD_compress],[zstd zstd.dll cygzstd.dll], [], [])
+ AC_DEFINE([HAVE_ZSTD], [1], [if true, zstd library is available])
+fi
+AC_MSG_CHECKING([whether libzstd library is available])
+AC_MSG_RESULT([${have_zstd}])
+````
+Note the the entry point (*ZSTD_compress*) is library dependent
+and is used to see if the library is available.
+
+#### Makefile.am
+
+It is assumed you have an HDF5 wrapper for zstd. If you want it
+to be built as part of the netcdf-c library then you need to
+add the following to *netcdf-c/plugins/Makefile.am*.
+````
+if HAVE_ZSTD
+noinst_LTLIBRARIES += libh5zstd.la
+libh5szip_la_SOURCES = H5Zzstd.c H5Zzstd.h
+endif
+````
+
+# Need our version of szip if libsz available and we are not using HDF5
+if HAVE_SZ
+noinst_LTLIBRARIES += libh5szip.la
+libh5szip_la_SOURCES = H5Zszip.c H5Zszip.h
+endif
+
+#### CMakeLists.txt
+In an analog to *configure.ac*, a block like
+this needs to be in *netcdf-c/CMakeLists.txt*.
+````
+FIND_PACKAGE(Zstd)
+set_std_filter(Zstd)
+````
+The FIND_PACKAGE requires a CMake module for the filter
+in the cmake/modules directory.
+The *set_std_filter* function is a macro.
+
+An entry in the file config.h.cmake.in will also be needed.
+````
+/* Define to 1 if zstd library available. */
+#cmakedefine HAVE_ZSTD 1
+````
+
+### Implementation Template
+As a template, here is the implementation for zstandard.
+It can be used as the template for adding other standard filters.
+It is currently located in *netcdf-d/libdispatch/dfilter.c*, but
+could be anywhere as indicated above.
+````
+#ifdef HAVE_ZSTD
+int
+nc_def_var_zstandard(int ncid, int varid, int level)
+{
+ int stat = NC_NOERR;
+ unsigned ulevel;
+
+ if((stat = nc_inq_filter_avail(ncid,H5Z_FILTER_ZSTD))) goto done;
+ /* Filter is available */
+ /* Level must be between -131072 and 22 on Zstandard v. 1.4.5 (~202009)
+ Earlier versions have fewer levels (especially fewer negative levels) */
+ if (level < -131072 || level > 22)
+ return NC_EINVAL;
+ ulevel = (unsigned) level; /* Keep bit pattern */
+ if((stat = nc_def_var_filter(ncid,varid,H5Z_FILTER_ZSTD,1,&ulevel))) goto done;
+done:
+ return stat;
+}
+
+int
+nc_inq_var_zstandard(int ncid, int varid, int* hasfilterp, int *levelp)
+{
+ int stat = NC_NOERR;
+ size_t nparams;
+ unsigned params = 0;
+ int hasfilter = 0;
+
+ if((stat = nc_inq_filter_avail(ncid,H5Z_FILTER_ZSTD))) goto done;
+ /* Filter is available */
+ /* Get filter info */
+ stat = nc_inq_var_filter_info(ncid,varid,H5Z_FILTER_ZSTD,&nparams,NULL);
+ if(stat == NC_ENOFILTER) {stat = NC_NOERR; hasfilter = 0; goto done;}
+ if(stat != NC_NOERR) goto done;
+ hasfilter = 1;
+ if(nparams != 1) {stat = NC_EFILTER; goto done;}
+ if((stat = nc_inq_var_filter_info(ncid,varid,H5Z_FILTER_ZSTD,&nparams,¶ms))) goto done;
+done:
+ if(levelp) *levelp = (int)params;
+ if(hasfilterp) *hasfilterp = hasfilter;
+ return stat;
+}
+#endif /*HAVE_ZSTD*/
+````
+
+# Point of Contact {#intern_poc}
+
+*Author*: Dennis Heimbigner
+*Email*: dmh at ucar dot edu
+*Initial Version*: 12/22/2021
+*Last Revised*: 01/25/2022
diff --git a/docs/nczarr.md b/docs/nczarr.md
index af10387c60..969a403228 100644
--- a/docs/nczarr.md
+++ b/docs/nczarr.md
@@ -481,9 +481,9 @@ collections — High-performance dataset datatypes](https://docs.python.org/2/li
[7] [XArray Zarr Encoding Specification](http://xarray.pydata.org/en/latest/internals.html#zarr-encoding-specification)
[8] [Dynamic Filter Loading](https://support.hdfgroup.org/HDF5/doc/Advanced/DynamicallyLoadedFilters/HDF5DynamicallyLoadedFilters.pdf)
[9] [Officially Registered Custom HDF5 Filters](https://portal.hdfgroup.org/display/support/Registered+Filter+Plugins)
-[10] [C-Blosc Compressor Implementation](https://github.com/Blosc/c-blosc)
-[11] [Conda-forge / packages / aws-sdk-cpp]
-(https://anaconda.org/conda-forge/aws-sdk-cpp)
+[10] [C-Blosc Compressor Implementation](https://github.com/Blosc/c-blosc)
+[11] [Conda-forge / packages / aws-sdk-cpp](https://anaconda.org/conda-forge/aws-sdk-cpp)
+[12] [GDAL Zarr](https://gdal.org/drivers/raster/zarr.html)
# Appendix A. Building NCZarr Support {#nczarr_build}
@@ -524,8 +524,7 @@ Note also that if S3 support is enabled, then you need to have a C++ compiler in
The necessary CMake flags are as follows (with defaults)
-1.
--DENABLE_NCZARR=off -- equivalent to the Automake _--disable-nczarr_ option.
+1. -DENABLE_NCZARR=off -- equivalent to the Automake _--disable-nczarr_ option.
2. -DENABLE_NCZARR_S3=off -- equivalent to the Automake _--enable-nczarr-s3_ option.
3. -DENABLE_NCZARR_S3_TESTS=off -- equivalent to the Automake _--enable-nczarr-s3-tests_ option.
@@ -562,7 +561,7 @@ Building this package from scratch has proven to be a formidable task.
This appears to be due to dependencies on very specific versions of,
for example, openssl.
-## **nix** Build
+## *\*nix\** Build
For linux, the following context works. Of course your mileage may vary.
* OS: ubuntu 21
@@ -682,7 +681,7 @@ Some of the relevant limits are as follows:
Note that the limit is defined in terms of bytes and not (Unicode) characters.
This affects the depth to which groups can be nested because the key encodes the full path name of a group.
-# Appendix D. Alternative Mechanisms for Accessing Remote Datasets
+# Appendix D. Alternative Mechanisms for Accessing Remote Datasets {#nczarr_altremote}
The NetCDF-C library contains an alternate mechanism for accessing traditional netcdf-4 files stored in Amazon S3: The byte-range mechanism.
The idea is to treat the remote data as if it was a big file.
@@ -706,7 +705,7 @@ Specifically, Thredds servers support such access using the HttpServer access me
https://thredds-test.unidata.ucar.edu/thredds/fileServer/irma/metar/files/METAR_20170910_0000.nc#bytes
````
-# Appendix E. AWS Selection Algorithms.
+# Appendix E. AWS Selection Algorithms. {#nczarr_awsselect}
If byterange support is enabled, the netcdf-c library will parse the files
````
@@ -764,7 +763,7 @@ Picking an access-key/secret-key pair is always determined
by the current active profile. To choose to not use keys
requires that the active profile must be "none".
-# Appendix F. NCZarr Version 1 Meta-Data Representation
+# Appendix F. NCZarr Version 1 Meta-Data Representation. {#nczarr_version1}
In NCZarr Version 1, the NCZarr specific metadata was represented using new objects rather than as keys in existing Zarr objects.
Due to conflicts with the Zarr specification, that format is deprecated in favor of the one described above.
@@ -779,6 +778,26 @@ The content of these objects is the same as the contents of the corresponding ke
* ''.nczarray <=> ''_NCZARR_ARRAY_''
* ''.nczattr <=> ''_NCZARR_ATTR_''
+# Appendix G. JSON Attribute Convention. {#nczarr_version1}
+
+An attribute may be encountered on read whose value when parsed
+by JSON is a dictionary. As a special conventions, the value
+converted to a string and stored as the value of the attribute
+and the type of the attribute is treated as char.
+
+When writing a character valued attribute, it's value is examined
+to see if it looks like a JSON dictionary (i.e. "{...}")
+and is parseable as JSON.
+If so, then the attribute value is treated as one long string,
+parsed as JSON, and stored in the .zattr file in JSON form.
+
+These conventions are intended to help support various
+attributes created by other packages where the attribute is a
+complex JSON dictionary. An example is the GDAL Driver
+convention [12]. The value is a complex
+JSON dictionary and it is desirable to both read and write that kind of
+information through the netcdf API.
+
# Point of Contact {#nczarr_poc}
__Author__: Dennis Heimbigner
|
|---|
|