Changelog 2023.7.18.dev1

Jul 18, 2023 • Anders Berkeman, Carl Drougge, and Sofia Hörberg

Highlights

• job.[json_]save() takes a background argument to let your method do other things while saving the file.
• Many usability tweaks in shell commands and board.

Board

• Support for sub-directories in the result directory.
• Zebra stripe text files and job output.
• Colour support in text files and job output. (By using accelerator.colour.)
• Job pages have ⇦ prev, next ⇨ and LATEST ⇉ links.
• Workdir lists can be sorted by clicking the headings.

csvimport

• labelsonfirstline is replaced by a more generic label_lines, so multi-line labels work (newlines are replaced by spaces).
• If labels is set, label_lines defaults to 0 (but you can of course set it if needed).
• Skipped lines (from labels, comment, skip_lines and skip_empty_lines) no longer affect the slicing of included lines. (Bad lines are however sliced together with the other included lines, if you set allow_bad.)

In your methods

• job.link_result() works with sub-directories (both source and destination).
• job.[json_]save() takes a background argument to let your method do other things while saving the file.
• job.[json_]save() returns an object with some methods and properties:
  • .load() (waits until saving is done)
  • .remove()
  • .filename
  • .path

Shell

• All commands allow --help (and sometimes more) without a config file.
• New ax status command, showing the currently running method and more interestingly the last error (like the board status page already did).
• New ax gc command to clean up workdirs. Currently only removes jobs that didn’t finish.
• ax job shows less output by default, adds --verbose to get it back.
• The job spec syntax is extended further, and now supports:
  • start with a jobid (e.g. dev-42), a method (e.g. csvimport) or an urd spec (:urdlist:[entry], e.g. :import/2023-07-18:).
  • ~ to go back one job of the same method.
  • ^ to follow .parent.
  • + to go forward one job of the same method.
  • ! to make ~ and + only consider current jobs.
  • < to go one jobid back (e.g. dev-42 ⇨ dev-41)
  • > to go one jobid forward (e.g. dev-42 ⇨ dev-43)
  • .item to follow item in argument jobs or datasets (use .jobs.item or .datasets.item to disambiguate).

  These can be chained freely, and accept counts where this makes sense. E.g. dataset_type~5.source^ or equivalently dataset_type~~~~~.source.parent. Remember to escape shell special characters (at most !^<>).

Urd

• Newlines (and everything else) are now allowed in captions.
• Urd can run without a config file, by specifying --listen.

And finally

• More tests.
• Various bug fixes / minor improvements of course.

Note: Python 2 is still supported, but will be unsupported soon.

Additional Resources

The Accelerator’s Homepage (exax.org)
The Accelerator on Github/exaxorg
The Accelerator on PyPI
Reference Manual