More octave fix #10
More octave fix #10
Conversation
"get_global_value ()" is defined in octave/variables.h In Octave 5.x, the previous collision with "get_global_value ()" was in the include chain: octave/parse.h -> octave/lex.h -> octave/input.h -> octave/hook-fcn.h -> octave/variables.h But in Octave 4.x, octave/parse.h -> octave/lex.h -> octave/input.h .. octave/hook-fcn.h -> octave/variables.h "octave/input.h" does not include "octave/hook-fcn.h" They have #define guards against multiple inclusions, so including "octave/variables.h" twice in Octave 5.x is safe.
|
Btw, there are a fair number of deprecated warnings against octave 4.x (I have 5.x on fedora but 4.x on raspbian) - I see you were on octave 3.x - do you worry about supporting that? Anyway - just questions that came up; don't necessarily mean I want to carry on, so this pull is as is - if I want to do those deprecation warnings that's another pull. |
|
We don't need to support very old octave versions. On desktop distributions they are typically much newer. On embedded build systems like openembedded and buildroot, we custom build to more recent versions. |
|
@flatmax personally, I think trying to support octave 3.x is a lost cause, considering that octave itself is playing catch-up with matlab. So perhaps one should just concentrate on what works with current octave (5.x) and if it is not too painful, one version back. I mean many of the warnings seems to be that you were copying/redefining octave types which has since become accessible. So this trend (of removing duplicate / colliding definitions) for current octave is likely to break compatibility with older octave. |
A few more updates. The list of dependencies are just new stuff not in my fedora laptop and my pi, not meant to be exhaustive.