Online Developer Meeting (2021-08-24)

From Octave
Revision as of 22:12, 24 August 2021 by Siko1056 (talk | contribs) (→‎Todays topics)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Todays topics[edit]

  • Meet and greet 5 minutes before meeting (audio testing).

Switch this meeting to more casual format[edit]

  • In the future notes of the talks will be added to this wiki in a more sparse fashion.

liboctave[edit]

  • Make more use of STL
  • Not all liboctave components can be replaced by STL containers, as the ND-indexing is not supported.
  • string_vector class probably replaceable by STL containers.

Tips for Octave C++ code writers[edit]

Such as packages, toolboxes, etc.

  • Prefer OCTAVE_LOCAL_BUFFER over heavy-weight Array object to store results of (Fortran-) library calls.
  • Prefer octave_value type variables over internal data structures.
    • Examples: Function handle class and Range class.

For package maintainers[edit]

octave namespace[edit]

  • jwe made many breaking changes on the default branch moving symbols in the octave namespace, to keep the namespaces tidy.

Deprecations[edit]

Trolling[edit]

  • In case of trolling behavior of users, please mark the thread, thus a Discourse admin can close the threat and mute the user.
    • DO NOT FEED THE TROLL: Any further answer or conversation is most likely a waste of emotions and time.
  • Any user violating the usage conditions https://octave.discourse.group/tos#heading--conditions will be muted or removed from the forum entirely.

Code sprints[edit]

Previous topics[edit]

The following items were not discussed. Just some links to progress on those items are displayed.

Octave 6.3 released[edit]

Octave 6.4 / 7 / 8[edit]

  • Discussion how to continue with the development.
    • jwe wants to introduce breaking changes (see below).
  • Octave should stick with the ~yearly major release cycle
    • Probably no Octave 6.4 (only if severe bugs occur).
    • Octave 7 (default branch)
      • Should no longer receive "very" breaking changes (e.g. String-class for Octave 8)
      • Has about 305 bugs fixed (not 100% reliable figure) and should not wait another year until many bigger outstanding changes happen (will be deferred to Octave 8)
      • Tentative plan:
        • November 2021 merge default to stable.
        • End of 2021 release of Octave 7
    • Octave 8 (new default branch)
      • No clear decision how to handle very breaking changes until November 2021 merge default to stable. Depends on future needs.

Octave 7 / 8[edit]

  • Function argument parsing (introduced in Matlab R2019b, rather new)
    • jwe will continue to work on this feature on Octave 7 (default branch)
    • If feature cannot be completed by the end of the year, it will be disabled on parser-level (error), and introduced in Octave 8 one year later.
  • New GUI command widget
    • Possible to introduce it as user opt-in in Octave 7
    • Pending issues:
      • Command-widget does not look like the previous one (textbox for command input)
      • jwe needs better looking widget (avoid developing a new one)
      • No possibility to run external applications (like emacs, pager), which is already partially broken now. Need to implement own paging strategy (scroll bars, etc.)
  • Deprecation of Octave operators
    • Improve Matlab compatibility
    • Some extensions make it painful to implement Matlab compatible command-style function calls? See also the discussion about this topic.
      • Remove rarely used extensions like "**" power.
      • Discussion about removal of "+=", "++", etc. No final decision made.
      • Often used extensions must probably stay (like "!" or "!=" used in place of "~" or "~=").

C++ shared pointer / liboctave[edit]

  • jwe had a look at Octave own reference counting
  • Wish to replace it with C++ shared pointers
  • Expert knowledge wanted! jwe opened a discussion thread
  • jwe identified "copy expensive" inefficiency about mxArray to octave_value conversion? Root of the trouble lies in historical handling of complex data? See this discussion.

See also[edit]