ChangeLog
+------------------------+
Coalispr-1.0.0 (May, 2026)
+------------------------+
'Parquet' release.

Pyarrow is an added requirement for running Coalispr now
    Pandas >= 3 employs pyarrow to improve dataframe handling through the open
    source parquet format, which is used here to reducee I/O & disk storage:

    Refactored binary storage from shelve/pickle to parquet  (**BNY**, **PARQ**).
    Removed ``resources/numeric.py`` needed for ``python.shelve`` with pandas_ >= 2.
    Store only merged bedgraph data; this removes redundancy and in-between data I/O.
    Removed editing of sample-names (because of fast data loading and counting).
    Removed backup to text as chosen binary format (parquet) is exchangeable.
    Made saving counts to tabbed text files (tsv) optional; defaults to parquet.
    Added a parquet-viewer to check outputs (in GUI and 'coalispr info -pV 1').
Changed multiprocessing to use 'spawn' to start child processes:
    Should facilitate usage on Windows or MacOS.
    Previous default ('fork') restricted multiprocessing to Linux/BSD.
    Refactored 'datastore, countbams, info'.

Manage memory usage in support of multiprocessing: 
    Large catalogs like chromosome indexes are no longer loaded into memory 
      but read from file by each sub-process.
    When needed, these catalogs are temporarily stored on disc.
Changed naming structure of created count and data files:
    API breakage: old count files are no longer processed by the program.
Tested and developed with Python 3.12; updated zenodo archive (v6).

+------------------------+
Coalispr-0.9.7 (Mar, 2026)
+------------------------+
Second public release.
Inclusion of a GUI (built with Tcl/Tk;  start with 'coalispr_gui') to compose
    and run the same commands as via the CLI.
Refactoring of counting, now based around 'collections.Counter' (replacing
    'pandas.DataFrame'). This gives a massive speed increase.
Multiprocessing used for counting and loading of bedgraph data (on linux).
Hand-cursor indicates interactive elements on interface of 'coalispr showgraphs'.
Bin-analysis more refined and possibly more informative.
'coalispr groupcompare' refactored.
Counts for reads linked to extrageneous DNA (CHRXTRA) are now kept apart so
    that these do not inflate genomic counts.
A category for additional chromosomal DNA (ADD_GDNA) - that first was taken as
    extra DNA - is now included to enable inclusion into genomic counts.
Tested and developed with Python 3.12; updated zenodo archive (v4).
Declared as "Production/Stable".

+------------------------+
Coalispr-0.7.8 (Oct, 2024)
+------------------------+
First public, bioRχiv, release.
H99 and Mouse data @ https://doi.org/10.5281/zenodo.12822543
Tested and developed with Python 3.9
