tskit-dev · hossam26644 · Apr 10, 2025
diff --git a/c/tskit/core.c b/c/tskit/core.c
@@ -168,7 +168,7 @@ tsk_strerror_internal(int err)
             break;
         case TSK_ERR_FILE_VERSION_TOO_OLD:
             ret = "tskit file version too old. Please upgrade using the "
-                  "'tskit upgrade' command. (TSK_ERR_FILE_VERSION_TOO_OLD)";
+                  "'tskit upgrade' command from tskit version<0.6.2. (TSK_ERR_FILE_VERSION_TOO_OLD)";
             break;
         case TSK_ERR_FILE_VERSION_TOO_NEW:
             ret = "tskit file version is too new for this instance. "

diff --git a/c/tskit/core.h b/c/tskit/core.h
@@ -283,7 +283,7 @@ A file could not be read because it is in the wrong format
 /**
 The file is in tskit format, but the version is too old for the
 library to read. The file should be upgraded to the latest version
-using the ``tskit upgrade`` command line utility.
+using the ``tskit upgrade`` command line utility from tskit version<0.6.2.
 */
 #define TSK_ERR_FILE_VERSION_TOO_OLD                                -101
 /**

diff --git a/docs/file-formats.md b/docs/file-formats.md
@@ -42,7 +42,7 @@ stored as well as the top-level metadata.
 ### Legacy Versions
 
 Tree sequence files written by older versions of tskit are not readable by
-newer versions of tskit. For major releases of tskit, `tskit upgrade`
+newer versions of tskit. For tskit releases<0.6.2, `tskit upgrade`
 will convert older tree sequence files to the latest version.
 
 

diff --git a/python/CHANGELOG.rst b/python/CHANGELOG.rst
@@ -17,6 +17,11 @@
 - Metadata.schema was returning a modified schema, this is fixed to return a copy of
   the original schema instead (:user:`benjeffery`, :issue:`3129`, :pr:`3130`)
 
+**Breaking Changes**
+
+- Legacy formats from msprime<0.6 (HDF5 formats) support is dropped. This includes the
+  support for ``tskit upgrade``  (:user:`hossam26644`, :issue:`2812`, :pr:`3138`)
+
 --------------------
 [0.6.1] - 2025-03-31
 --------------------
@@ -142,7 +147,7 @@
   pass a subset of nodes, so subtrees can be visually collapsed. Additionally, an option
   ``pack_untracked_polytomies`` allows large polytomies involving untracked samples to
   be summarised as a dotted line (:user:`hyanwong`, :issue:`3011` :pr:`3010`, :pr:`3012`)
-  
+
 - Added a ``title`` parameter to ``.draw_svg()`` methods (:user:`hyanwong`, :pr:`3015`)
 
 - Add comma separation to all display numbers. (:user:`benjeffery`, :issue:`3017`, :pr:`3018`)

diff --git a/python/_tskitmodule.c b/python/_tskitmodule.c
@@ -223,7 +223,7 @@ handle_library_error(int err)
     const char *not_kas_format_msg
         = "File not in kastore format. Either the file is corrupt or it is not a "
           "tskit tree sequence file. It may be a legacy HDF file upgradable with "
-          "`tskit upgrade` or a compressed tree sequence file that can be decompressed "
+          "`tskit upgrade` from tskit version<0.6.2 or a compressed tree sequence file that can be decompressed "
           "with `tszip`.";
     const char *ibd_pairs_not_stored_msg
         = "Sample pairs are not stored by default "

diff --git a/python/requirements/CI-complete/requirements.txt b/python/requirements/CI-complete/requirements.txt
@@ -1,7 +1,6 @@
 biopython==1.85
 coverage==7.7.0
 dendropy==5.0.1
-h5py==3.13.0
 kastore==0.3.3
 lshmm==0.0.8
 msgpack==1.1.0

diff --git a/python/requirements/CI-tests-conda/requirements.txt b/python/requirements/CI-tests-conda/requirements.txt
@@ -1,4 +1,3 @@
 msprime==1.3.3
 tszip==0.2.5
-h5py==3.13.0
 zarr<3
diff --git a/python/requirements/development.txt b/python/requirements/development.txt
@@ -6,7 +6,6 @@ codecov
 coverage
 dendropy
 flake8
-h5py>=2.6.0
 jsonschema>=3.0.0
 jupyter-book>=0.12.1
 kastore
@@ -22,7 +21,7 @@ numpy
 packaging
 portion
 pre-commit
-pyparsing 
+pyparsing
 pysam
 pytest
 pytest-cov

diff --git a/python/requirements/development.yml b/python/requirements/development.yml
@@ -13,7 +13,6 @@ dependencies:
   - dendropy
   - doxygen
   - flake8
-  - h5py>=2.6.0
   - jsonschema>=3.0.0
   - jupyter-book>=0.12.1
   - kastore

diff --git a/python/tests/test_cli.py b/python/tests/test_cli.py
@@ -1,6 +1,6 @@
 # MIT License
 #
-# Copyright (c) 2018-2024 Tskit Developers
+# Copyright (c) 2018-2025 Tskit Developers
 # Copyright (c) 2017 University of Oxford
 #
 # Permission is hereby granted, free of charge, to any person obtaining a copy
@@ -30,7 +30,6 @@
 import unittest
 from unittest import mock
 
-import h5py
 import msprime
 import pytest
 
@@ -311,16 +310,6 @@ def test_vcf_allow_position_zero(self, flags, expected):
         assert args.tree_sequence == tree_sequence
         assert args.allow_position_zero == expected
 
-    def test_upgrade_default_values(self):
-        parser = cli.get_tskit_parser()
-        cmd = "upgrade"
-        source = "in.trees"
-        destination = "out.trees"
-        args = parser.parse_args([cmd, source, destination])
-        assert args.source == source
-        assert args.destination == destination
-        assert not args.remove_duplicate_positions
-
     def test_info_default_values(self):
         parser = cli.get_tskit_parser()
         cmd = "info"
@@ -655,68 +644,3 @@ def test_migrations(self):
 
     def test_provenances(self):
         self.verify("provenances")
-
-
-class TestUpgrade(TestCli):
-    """
-    Tests the results of the upgrade operation to ensure they are
-    correct.
-    """
-
-    def setUp(self):
-        fd, self.legacy_file_name = tempfile.mkstemp(prefix="msp_cli", suffix=".trees")
-        os.close(fd)
-        fd, self.current_file_name = tempfile.mkstemp(prefix="msp_cli", suffix=".trees")
-        os.close(fd)
-
-    def tearDown(self):
-        os.unlink(self.legacy_file_name)
-        os.unlink(self.current_file_name)
-
-    def test_conversion(self):
-        ts1 = msprime.simulate(10)
-        for version in [2, 3]:
-            tskit.dump_legacy(ts1, self.legacy_file_name, version=version)
-            stdout, stderr = capture_output(
-                cli.tskit_main,
-                ["upgrade", self.legacy_file_name, self.current_file_name],
-            )
-            ts2 = tskit.load(self.current_file_name)
-            assert stdout == ""
-            assert stderr == ""
-            # Quick checks to ensure we have the right tree sequence.
-            # More thorough checks are done elsewhere.
-            assert ts1.get_sample_size() == ts2.get_sample_size()
-            assert ts1.num_edges == ts2.num_edges
-            assert ts1.get_num_trees() == ts2.get_num_trees()
-
-    def test_duplicate_positions(self):
-        ts = msprime.simulate(10, mutation_rate=10)
-        for version in [2, 3]:
-            tskit.dump_legacy(ts, self.legacy_file_name, version=version)
-            root = h5py.File(self.legacy_file_name, "r+")
-            root["mutations/position"][:] = 0
-            root.close()
-            stdout, stderr = capture_output(
-                cli.tskit_main,
-                ["upgrade", "-d", self.legacy_file_name, self.current_file_name],
-            )
-            assert stdout == ""
-            tsp = tskit.load(self.current_file_name)
-            assert tsp.sample_size == ts.sample_size
-            assert tsp.num_sites == 1
-
-    def test_duplicate_positions_error(self):
-        ts = msprime.simulate(10, mutation_rate=10)
-        for version in [2, 3]:
-            tskit.dump_legacy(ts, self.legacy_file_name, version=version)
-            root = h5py.File(self.legacy_file_name, "r+")
-            root["mutations/position"][:] = 0
-            root.close()
-            with mock.patch("sys.exit", side_effect=TestException) as mocked_exit:
-                with pytest.raises(TestException):
-                    capture_output(
-                        cli.tskit_main,
-                        ["upgrade", self.legacy_file_name, self.current_file_name],
-                    )
-                assert mocked_exit.call_count == 1