Online Developer Meeting (2022-08-23): Difference between revisions

From Octave
Jump to navigation Jump to search
(→‎Previous topics: Move in updated notes from previous meeting)
Tags: Mobile edit Mobile web edit
Line 7: Line 7:


* GSoC progress
* GSoC progress
** '''Progress was presented by available mentors. Generally good. Some discussion on openlibm, accepted that negative results are also valuable.'''
** Progress was presented by available mentors. Projects are on track.
* VM progress?
** Openlibm project concluded that the approach doesn't work. Valuable negative result!
** '''Petter presented updates on his VM.'''
** libtiff project will probably implement the classdef as an .m file. (At least the constructor, which is hard/impossible to implement in C++ code. It is not a "normal" DLDFCN). Will probably move from dld to core (libtiff dependency).
** '''jwe prefers incorporating that into the default branch of Octave as soon as practical, and let it remain off by default unless the user wants to enable it with a configure switch or a runtime switch'''
* When should a new function go to core or load dynamically?
** '''This would be formed into an API for a generic VM approach to Octave that can be used for other VMs such as the Jitter one, etc'''
** If the dependencies are already linked into Octave core -> No big benefit to have it dynamically loaded. More sensible to add in core.
** If function loads a big (new) dependency, it might make sense to load it "on demand". --> dldfcn
* VM progress:
** Petter presented updates on his VM. Good progress. But still a lot to do.
** will incrementally move from fork to default branch as an experimental feature with a configure option to disable it.
** will most likely not be active by default in Octave 8 (either not compiled or de-activated on run-time)
** won't stop the option for alternative VM implementations. Might help to define an "API" for other VM implementations (like Jitter, etc).
* Release process for Octave 7.2:
* Release process for Octave 7.2:
** About 10 (actual) changes since the last release candidate (about 6 weeks ago). Should we make a second release candidate?
** About 10 (actual) changes since the last release candidate (about 6 weeks ago). Should we make a second release candidate?
*** No new release candidate. Proceed with a release after some final preparations (e.g., NEWS file).
** Not much feedback for the first release candidate. Should we skip making release candidates entirely and directly make releases? Should we announce them more prominently (e.g., on the web page)?
** Not much feedback for the first release candidate. Should we skip making release candidates entirely and directly make releases? Should we announce them more prominently (e.g., on the web page)?
* Classdef
*** Call for ideas.
** '''Technical discussion happened on things specific to classdefs'''
* Octave packages progress update
* Octave packages progress update
* Managing diversity of libraries (libm, libc etc)
* Managing diversity of libraries (libm, libc etc)

Revision as of 23:01, 26 July 2022

Today's topics

Previous topics

  • GSoC progress
    • Progress was presented by available mentors. Projects are on track.
    • Openlibm project concluded that the approach doesn't work. Valuable negative result!
    • libtiff project will probably implement the classdef as an .m file. (At least the constructor, which is hard/impossible to implement in C++ code. It is not a "normal" DLDFCN). Will probably move from dld to core (libtiff dependency).
  • When should a new function go to core or load dynamically?
    • If the dependencies are already linked into Octave core -> No big benefit to have it dynamically loaded. More sensible to add in core.
    • If function loads a big (new) dependency, it might make sense to load it "on demand". --> dldfcn
  • VM progress:
    • Petter presented updates on his VM. Good progress. But still a lot to do.
    • will incrementally move from fork to default branch as an experimental feature with a configure option to disable it.
    • will most likely not be active by default in Octave 8 (either not compiled or de-activated on run-time)
    • won't stop the option for alternative VM implementations. Might help to define an "API" for other VM implementations (like Jitter, etc).
  • Release process for Octave 7.2:
    • About 10 (actual) changes since the last release candidate (about 6 weeks ago). Should we make a second release candidate?
      • No new release candidate. Proceed with a release after some final preparations (e.g., NEWS file).
    • Not much feedback for the first release candidate. Should we skip making release candidates entirely and directly make releases? Should we announce them more prominently (e.g., on the web page)?
      • Call for ideas.
  • Octave packages progress update
  • Managing diversity of libraries (libm, libc etc)
  • Octave 7.3 and 8 timeline?

See also