On Sun, 14 Apr 2013 10:54:44 +0300 "daniel.za...@samsung.com" <daniel.za...@samsung.com> said:
> On 04/14/2013 10:27 AM, Carsten Haitzler (The Rasterman) wrote: > > On Sun, 14 Apr 2013 10:06:47 +0300 "daniel.za...@samsung.com" > > <daniel.za...@samsung.com> said: > > > >> Hi Raster, > >> > >> If I understand correctly, you would like to add some layer above Eo > >> that would simplify writing code. I don't think people suffered enough > >> to "hide" Eo ;-) > >> As you said, a lot of details have to be checked, like for example how > >> do you call these functions? with the previous writing way or with a > >> shorten one? > >> Same thing with eo_data_get and eo_do_super. > > you call like now: eo_do(obj, methoda(x,y,z), methodb()); > > etc. we are just talking of how to cut down the boilerplate and reduce > > errors, maing it easier. > If we take your example, what is methoda? font_source_set or > eo_font_font_source_set? With the first one, we must be sure there is > not the same function name in another class (for the moment, we have > this kind of issue). With the second one, the developer has to guess the > function name. probably 2nd one - but with the 2nd the public header is generated and that is the canonical/actual api. > > > >> If we really go on this way, I would suggest to change the manner get > >> functions return the values. Maybe something like Python. But here, same > >> as before, have to check all the corners. > >> > >> Imo, we don't need to re-invent a new language that will take time to > >> adapt to the existing code and may not be approved by all the world (as > >> with Eo). I think a tool is preferable so I would go to something like > >> Cedric proposed. And DO NOT THINK I agree with Cedric because we are > >> both French (and you not) !!! > > not thinking new lang. just thinking easier way to declare all the eo > > boilerplate and maintain it. its really just a better cpp before cpp. i > > don't see why not to be able to just write my c code method content right > > there and then. otherwise this generator has to not just generate code by > > EDIT existing code that you will have edited after intial generation as > > opposed to just generate the surrounding boilerplate c ANd insert the code > > you just wrote as part of it. > I understand your point. But the real question is which issue we want to > solve. If we just want to add/remove Eo functions, a simple tool is easy > to write, sufficient and could be easily integrated into a future EFL > IDE. Updating the code is the hardest part but a simple C parser can > make the work easy (imo). > If we want to adapt the syntax to the developer mindset, it is another > story that will take more time with more virulent discussions (hehe, as > usual in open-source). But it could be a better long-term solution. I > really don't know. > > Maybe we need both tools. The simple tool for now, with possibility to > extend meta-data files to accept code in the future. And in parallel, > thinking and designing a new way to develop on EFL. > > > >> JackDanielZ, your favorite larbin > >> > >> On 04/11/2013 02:09 PM, Carsten Haitzler (The Rasterman) wrote: > >>> On Thu, 11 Apr 2013 14:06:04 +0900 Cedric BAIL <cedric.b...@free.fr> said: > >>> > >>>> Hello everyone, > >>>> > >>>> One of the annoying things with Eo is that we now need to update a few > >>>> more place than before when we add a new function. We have to do : > >>>> - Add enum at the proper place in header > >>>> - Add macro for the function call in header > >>>> - Add function with proper argument in the C file > >>>> - Add declaration at the right place in the object table in the C file > >>>> - Add description at the right place in the description table in the C > >>>> file If necessary, we also have to do > >>>> - Add legacy prototype function in header > >>>> - Add legacy wrapper function in c file > >>>> > >>>> Of course when you create a new object you have even more boilerplate > >>>> to write. In fact, most of it can be inferred if we did have a > >>>> properly detailed prototype. So I would like to propose a tool, eo-gen > >>>> that would take a "json" like format (see a possible example there: > >>>> https://phab.enlightenment.org/P17 ). From this description file, it > >>>> will generate or update the .c and .h file. This json file would be > >>>> included into the distributed header of our library, so it could be > >>>> used by other tools. > >>>> > >>>> To properly update the .c/.h files, it will search for a know pattern > >>>> in the comment and use that to infer what is going on. Once using > >>>> eo-gen, it will not be possible to remove those comment without > >>>> breaking the capability to update the source. > >>>> > >>>> I know that some people have also been thinking about this problem, so > >>>> let's try to find a common solution there. > >>> frankly i would prefer a syntax that works with the programmers > >>> mindset/mode of working better: > >>> > >>> https://phab.enlightenment.org/P18 > >>> > >>> and this leads onto the next logical step: > >>> > >>> https://phab.enlightenment.org/P19 > >>> > >>> ie you can even place the code for methods right into the template and > >>> they get output along with the boilerplate. (in this we make objd always > >>> be the object private data etc. - sure some things not handled/explained > >>> - more a position on syntax style and where it leads). > >>> > >>>> -- > >>>> Cedric BAIL > >>>> > >>>> ------------------------------------------------------------------------------ > >>>> Precog is a next-generation analytics platform capable of advanced > >>>> analytics on semi-structured data. The platform includes APIs for > >>>> building apps and a phenomenal toolset for data science. Developers can > >>>> use our toolset for easy data analysis & visualization. Get a free > >>>> account! http://www2.precog.com/precogplatform/slashdotnewsletter > >>>> _______________________________________________ > >>>> 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" -------------- The Rasterman (Carsten Haitzler) ras...@rasterman.com ------------------------------------------------------------------------------ Precog is a next-generation analytics platform capable of advanced analytics on semi-structured data. The platform includes APIs for building apps and a phenomenal toolset for data science. Developers can use our toolset for easy data analysis & visualization. Get a free account! http://www2.precog.com/precogplatform/slashdotnewsletter _______________________________________________ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
