notes.txt

SMALL BUGS
- get_execution_paths() is returning 2 paths, when it should return 1
- annoying final UNKNOWN token end of cmdstmt
- resolve aliases doing weird shit  (hisat2) -x '$$reference_genome.index.fields.path'
- 'skeleton' mode should autotemplate tool inputs/arguments if they reference a process input
- filename types: 
    - always retain
    - janis_pipelines - CutAdaptBase_2
    - need pre-script for filename types
    - determine default
        - either "generated" or some form like ${output_prefix}-R1.fastq.gz
    - reference prescript redefinition in script 
- file type inputs - optional with default pointing to another tool input:
    - janis_pipelines - FastQCBase (extended):
      def read1 = read1.simpleName != params.NULL_VALUE ? read1 : ${reads1}
    - janis_pipelines - FastQCBase (regular):
      missing, but should be present! 



REGULAR MODE BUGS
0.  JANIS -> NEXTFLOW (BwaAlignment())
    - GETFASTQC_ADAPTERS should have 'read1_fastqc_datafile' as input
    - GETFASTQC_ADAPTERS should have 'read2_fastqc_datafile' as input

1.  CWL -> NEXTFLOW
    runtime inputs (runtime.cpu etc) should be kept in skeleton / regular mode

6.  ::GALAXY FEATURE::
    for options which appear in the templated command, mark these as mandatory.
    exceptions:
        - value is a galaxy param which we know is optional (in conditional param etc, or the param is marked optional)


TODO: TEST SUITE 100%
- nextflow
    - baseName vs simpleName?
- wdl 
    - miniwdl can't parse some WDL formats
    - wdl ingest: need to injest local variables (bwa.wdl Int totalSortThreads = select_first([sortThreads, estimatedSortThreads]))
    - mismatch secondary types
    - scatter on multiple inputs
    - nameroot operator - ask richard

NICETIES:
- galaxy 
    - make it faster
    - output collection especially with redirect is scuffed
    - picard_MarkDuplicates.xml: REMOVE_DUPLICATES='${remove_duplicates}' causing issues

TODO: RELEASE
- test cli
    - --galaxy-gen-images cmdline option etc
- all test-suites 
- all packages segregated as much as possible (ensure legacy compatibility)
- all packages build from pyproject.toml 
- ensure datatypes correctly segregated

FUTURE FEATURES
- prune_workflow() 
    - move static values to step.sources
- nextflow
    - Null values
        cwl subworkflow_test wf 
        looks weird: 
        OPTIONAL2(
            [].collect{ file(it) },         // in_file_arr
            file( params.NULL_VALUE ),      // in_file
            params.in_string_arr_optional,  // in_string_arr
            params.in_int_optional,         // in_int
            params.in_string_optional       // in_string
        )
- galaxy
    - annotation
        - move InputExtractor.add_uncaptured_inputs() to an annotation class.  
        - need to ensure not picking up gxparams which are only control params
    - repeat params (& conditional params?)
        - add support
        - cutadapt adapters argument
        - cutadapt library.type single?
    - command
        - which one is the main software command? 
        - future: create step for each software command, for non-linux software (mkdir, cp etc)?
    - select & when detection
        - mutually exclusive groups (cli args)
        - what to add to cmdstr 
        - .when() in janis? 
        - how will this appear in nextflow for tool translations? 
        - can we autoremove command compoennts for workflow translations?
            - would be only if all instances of the tool use have the same selects. 
            - others would need to be the same as tool translation. 
    - repeat block evaluation?