Enhancements and cleanup

TODO.md

@@ -11,45 +11,12 @@ SPDX-License-Identifier: Apache-2.0
 
 Support fo scanning the Elixir ecosystem needs to be added.
 
+* `audit` should seamlessly audit Elixir applications and libraries for AtomVM
+compatibility.
 * `ecosystem` command should have an --elixir (or --ex) option for creating an Elixir
 ecosystem data-set.
-
-### Change data structure for stored version info
-
-The current version data is stored as a binary string for tags and branches,
-this is brittle when comparing versions in the `update` command, and will cause
-noticeable errors if any AtomVM release versions use double digits for major,
-minor or patch levels. The storage format should be migrated to tuples.
-
-#### Tagged releases
-
-```erlang
-{
-    Major :: non_neg_integer(),
-    Minor :: non_neg_integer(),
-    Patch :: non_neg_integer()
-}
-```
-
-#### Branches
-
-##### main
-
-```erlang
-{
-    unreleased,
-    main
-}
-```
-
-##### `release-X.X`
-
-```erlang
-{
-    unreleased,
-    {release, Major :: non_neg_integer(), Minor :: non_neg_integer()}
-}
-```
+* `filter` command needs to be updated to work on Elixir ecosystem data with an
+--elixir (or --ex) option
 
 ## Should have
 
@@ -63,38 +30,6 @@ bare call like length(X) to resolve to the local function instead of the BIF,
 but the code will still count it as an OTP call, misclassifying user-defined
 functions and skewing scan results.
 
-### `supported` modules
-
-The `supported` command should print a list of all AtomVM modules if the `-m`
-or `--module` option is given without a module name.
-
-### Finer platform support tracking
-
-The tracking of platform support is not perfect. Some modules, like `network`
-end up being assigned too broad of platform support, in this case including
-`generic_unix` in the supported platforms, due to modules being assigned by
-library so all modules in `avm_network` are reported as supported by `esp32`,
-`generic_unix`, and `rp2` platforms. The `network` module is not supported on
-`generic_unix`, only `esp32` and `rp2`, but to track these exceptions specific
-filtering rules will be needed.
-
-The version added data should be tracked per-platform. For example `i2c` and
-`spi` added support for `rp2` and `stm32` platforms in version 0.7.0, while
-`esp32` had support in 0.5.0 and the `supported` command reports support for
-all platforms, and reports 0.5.0 as the release these functions were introduced
-(inaccurate for `rp2` and `stm32` platforms). The data storage format needs to
-be altered to track support for each platform, with `all` only requiring a
-single entry with the version.
-
-#### Track when modules or functions are deprecated and removed
-
-The supported functions data should track when modules are deprecated, and
-also when they are removed. Some new data structure will need to be devised,
-either a new field entirely, or expanding the `since` which will also be
-holding platform specific release introductions. The deprecation and removal
-releases may need to be hard-coded into the application, as these are rare, and
-parsing doc strings could potentially lead to false positives.
-
 ### Add support for adding (and reporting) downstream drivers and libraries
 
 The `update` command should have an option for adding downstream drivers or
@@ -108,20 +43,6 @@ supported functions as bare atoms, and downstream libraries as
 AtomVM version parameters, defaulting to `all` platforms and unknown for the
 AtomVM release.
 
-## Would be nice
-
-### Use logger with configurable levels
-
-Logger should be used instead of `io:format/2` for log messages. A configurable
-log file should be used, defaulting to a log file in the users cache directory
-that is overwritten on each run. The log level should be configurable, as well
-as the option for changing the log file name and location.
-
-#### Refactor error handling and logging
-
-Errors should be refactored to return atom() "reasons", and the conversion to
-log messages should be handled by dispatch to an error logger.
-
 ### Reusable APIs
 
 Most modules should be refactored to better separate logic and IO (reporting

generate_fun_data.sh

@@ -255,13 +255,13 @@ for PRE_TAG in ${PRE_TAGS_LIST}; do
     fi
 done
 
-# Sort tags by version (newest first for processing)
+# Sort tags by version (oldest first for processing)
 # Use git's sort if available, otherwise fall back to lexical sort
 if git tag -l --sort=v:refname "v*.*.*" 2>/dev/null | head -1 >/dev/null 2>&1; then
-    sort -rV "${TMP_TAG_FILE}" > "${TMP_TAG_FILE}.sorted" 2>/dev/null || sort -r "${TMP_TAG_FILE}" > "${TMP_TAG_FILE}.sorted"
+    sort -V "${TMP_TAG_FILE}" > "${TMP_TAG_FILE}.sorted" 2>/dev/null || sort "${TMP_TAG_FILE}" > "${TMP_TAG_FILE}.sorted"
 else
     # Lexical sort is acceptable for final ordering
-    sort -r "${TMP_TAG_FILE}" > "${TMP_TAG_FILE}.sorted"
+    sort "${TMP_TAG_FILE}" > "${TMP_TAG_FILE}.sorted"
 fi
 mv "${TMP_TAG_FILE}.sorted" "${TMP_TAG_FILE}"
 

include/ecosystem.hrl

@@ -10,5 +10,7 @@
 -define(ECOSYSTEM_HRL, true).
 
 -define(ECOSYSTEM_STATE, "beam_ecosystem.bin").
+-define(GITHUB_MAX_PER_QUERY, 1000).
+-define(HEX_PER_PAGE, 100).
 
 -endif.

include/function.hrl

@@ -0,0 +1,28 @@
+%%
+%% Copyright (c) 2026 Winford (UncleGrumpy) <winford@object.stream>
+%%
+%% This is part of atomvm_spectrometer
+%%
+%% SPDX-FileCopyrightText: 2026 Winford (UncleGrumpy)  <winford@object.stream>
+%% SPDX-License-Identifier: Apache-2.0
+
+-ifndef(FUNCTION_HRL).
+-define(FUNCTION_HRL, true).
+
+-type version_tuple() ::
+    {Major :: non_neg_integer(), Minor :: non_neg_integer(), Patch :: non_neg_integer()}
+    | {release, Major :: non_neg_integer(), Minor :: non_neg_integer()}
+    | {main, Major :: non_neg_integer(), Minor :: non_neg_integer()}.
+
+-type platform() :: esp32 | emscripten | generic_unix | rp2 | stm32.
+
+-type platform_versions() :: #{platform() | all => version_tuple()}.
+
+-record(function, {
+    name :: binary(),
+    arity :: non_neg_integer() | all,
+    since_map :: platform_versions(),
+    removed :: version_tuple() | undefined
+}).
+
+-endif.