On Fri, 13 Mar 2020 13:48:14 +0100 Stefan Schmidt <ste...@datenfreihafen.org>
said:
actually it seems i was wrong. marcel pointed this out - the evas 3d api was
always beta so we can nuke it entirely... no need fro empty functions to keep
abi :)
> Hello.
>
> On 13.03.20 12:26, Carsten Haitzler (The Rasterman) wrote:
> > On Fri, 13 Mar 2020 16:22:51 +0900 Hermet Park <hermetp...@gmail.com> said:
> >
> >> Leave apis empty? What do you mean by here APIs?
> >> I have no idea how this helps for preventing abi break. Though it might
> >> help to prevent api break but either way nothing helps.
> >
> > keep the legacy functions but with no content
> >
> > EAPI void evas_xxx_set(Evas_Object *obj, Eina_Bool x) { }
> > EAPI Evas_Object *evas_xxx_add(Evas *e) { return NULL; }
> >
> > etc. :) so any app that happens to call the api's can execute, just the
> > api's do nothing. no need to have .eo files - not even the classes or
> > actual object types. just enough empty functions to keep the abi from
> > breaking, but make the object totally non-functional. :) the object already
> > was a "soft guarantee". that means you needed a working gl engine for it to
> > work. if you work on software it'd only work if you also happened to have
> > osmesa also installed and this was dlopen()'d at runtime so it could
> > actually make this object not function at runtime. the above keeps the abi
> > there and just makes it never work no matter what the host setup was. :)
> > you could put in some ERR("Deprecated") too in the funcs if you wanted as
> > well.. :) so an executable that used these api's will start without symbol
> > errors. just this object will cease to work.
>
> If this should be changed for the upcoming 1.24 I would like to get this
> in rather sooner than later. To give it enough testing.
>
> ABI checker will also help us to find problems we might miss during
> normal testing.
>
> regards
> Stefan Schmidt
>
>
> _______________________________________________
> enlightenment-devel mailing list
> enlightenment-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
>
--
------------- Codito, ergo sum - "I code, therefore I am" --------------
Carsten Haitzler - ras...@rasterman.com
_______________________________________________
enlightenment-devel mailing list
enlightenment-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
