From: Carsten H. (T. R. <ra...@ra...> - 2010-10-19 23:27:26
|
On Tue, 19 Oct 2010 13:34:17 -0200 Raphael Kubo da Costa <ku...@pr...> said: > At Tue, 19 Oct 2010 16:31:38 +0200 (CEST), > Vincent Torri wrote: > > On Tue, 19 Oct 2010, Raphael Kubo da Costa wrote: > > > > > Hey all, > > > > > > The installed Eina.h includes the files in the eina/ subdirectory via > > > #include "foo.h" instead of #include "eina/foo.h" or #include > > > <eina/foo.h>. > > > > > > Are people always expected to -I both eina-1's path and eina-1/eina? I > > > thought #includ'ing <Eina.h> and -I'ing $PATH_TO_EINA-1 was enough, > > > and the rest was details. > > > > use what > > > > pkg-config --cflags eina > > > > returns, and include *only* Eina.h. > > In this specific project, I was trying to stick to CMake and keep > pkgconfig usage to a minimum. It is possible to follow your > suggestion, however I was thinking of GStreamer, for example, which > has a gst.h that includes headers in subdirectories, but whose .pc > includes only the base directory. aaah and this is where things clash. we TOTALLY rely on pkg-config as part of the solution. if you try to not use it, you are fighting against EFL. literally we DEPEND on pkg-config to be able to, in future, change include dirs, paths and other such options. you're driving down the wrong side of the road against traffic here. > Doesn't it make more sense to keep the eina/ subdirectory "away from > the public"? headers installed by their very nature are a totally public affair. its the public advertised interface to your library. it specifies how to use it for other apps/libs. if there is 1 bit of a library that is by far the MOST important bit to be clean, near, tidy and correct - it's the installed headers. -- ------------- Codito, ergo sum - "I code, therefore I am" -------------- The Rasterman (Carsten Haitzler) ra...@ra... |