Gustavo wrote: >>> This code is part of our effort to unify common code into base libs, >>> where its worth to avoid duplicating code over and over again. >>> >>> With that in Evas, we can now expose it in Edje and use it in Guarana >>> and Elementary and possible any other toolkit. It uses Evas_Object's >>> size hints infrastructure, but can be extended to have its own >>> properties if one wish. >>> >>> Attached is a test case and example. You can try to hack your own >>> fancy layout function, it's it's generic and useful to others, add it >>> to evas as well (Like layout using a Bezier or any other path). >>> >>> People implementing Ecore Evas can now use layout "stack" and >>> ecore_evas_object_associate() to provide the "main window" concept, >>> just add your background and then your object to the evas_object_box >>> and associate this box to the window. If you want your background to >>> scale in both directions, just set it's aligns to -1.0 and -1.0. >>> >>> >>> >> This is the kind of thing that I hoped evas would avoid tending to, and >> yet both raster and gustavo seem prone to want to get into. :( >> I know it's tempting to add more and more specialized layout/widgetry >> 'objects' to the core evas lib/api, but I think it's a big mistake for evas >> in the long run. >> >> It's not hard to start extending evas *and* edje to allow for custom >> objects, >> even if at first all such object modules need to access evas internals. >> Adding >> these kinds of things, whether text-grid objs, or box objs, or table objs, >> etc.. >> directly into the evas lib api is likely to become 'not a good idea'. >> > > this requires no access to internals, it's just a helper to avoid > spreading this code everywhere (e_box, guarana, elementary, etk...). > > the point is, raster always planned to have dynamic layout options in > Edje, so it's possible to specify in Edje if you want a stack, > vertical or horizontal layout and in code you add the objects. To add > this we need this kind of code. We could create yet another library > (like split esmart into edje independent and dependent code) and add > this there, make edje dependent on it, but it will be lot of work to > get a simple and useful code in. > > Since evas already have internal libs in it, at some point we can > create a libevas_utils and users can optionally link to it. That way > we keep things in a common place and avoid extra overhead. > > Our point is to do now what others should have done already: unify > stuff. Size hints, box, editable text block and these stuff should go > in Evas and make it easy to share stuff and easier to use EFL. I > always hear, at least from ETK, that some things were re-done in ETK > just to work around some problems in Evas or Edje, I consider that > wrong, these stuff should be fixed there, and that's what we're doing. > >
Ok, let me pose this this from a different view.. let's not worry about whether or not such 'layout' or 'widgetry' or 'special-purpose' or whatever kinds of objects should be added to the evas lib *api* -- I don't think they should in some cases, but that's partly incidental here. There will be a need for such kinds of objects, and others... things like custom self-rendered objs (possibly relying on other libs and apsects that those libs allow) that are best implemented at the 'evas level' for efficiency reasons and whatnot. It's just a matter of where their implementation should be, how to realize them.. and how to expose their apis. So, let's discuss the underlying issue that I think is what should be addressed here - namely, that of evas' "extensibility" at least as far as object types goes. Would one want things like "object modules", or libs of them, for giving evas that kind of of flexibility and extensibility at a low enough level? Well, one could certainly obtain the kinds of objs you want here, quite trivially really, and have them implemented by some lib.. even by edje itself say. And you could have things like a 'cairo' or an 'svg' object (which for certain engines could be done more 'directly' then via having these render to an argb buffer), and you could have say an 'ogre3d' object, or maybe a 'gstreamer' obj, or a video obj, or many, many things... and not have to worry about these being 'added' to the evas lib's api, or imposing a hard dependence on them by evas.. and yet, they can be implemented at the 'evas' level (possibly having them need access to evas internals at first, as I quickly outlined in an erlier email, and maybe further refined into a more abstract mechanism later if desired). So, as a question: Would this kind of extensibility for evas be something desired? What concrete things could you do with this? Is it a 'better' way to 'add' special/custom objs to evas then always worrying about whether or not to extend the evas lib's api with this or that 'one' useful obj? ____________________________________________________________ Click for online loan, fast & no lender fee, approval today http://thirdpartyoffers.juno.com/TGL2141/fc/Ioyw6i3m3WMEcU9j63sBFezjnV2gieGKKl9VFja0RSIJUWn2pEUyPS/ ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
