Re: [paragui-users] gradient fix....

[Teunis Peters]

On Fri, 1 Nov 2002, Dinand Vanvelzen wrote:

> Seperate libs would be a greater hassle for the end user i think,.. a define
> is easy to (un)comment
> always a pain in the ass wo get all includes and shit right for seperate
> libs, thats why i myself prefer #define switches,..a little more work to
> implement maybe but will save more work later on in my opninion.

warning: really rambly message follows.  Can't think of how to edit it so
my vote's at the end of it *wry grin*....
- - -

The biggest problem with a single header for everything is .... umm...
severly overloaded compiler problem.  I don't know about y'all out there
outside of gcc-land, but gcc's happier with lots of little headers as near
as I can tell....

BUT - a single header that includes paragui headers as needed makes nice
sense.  Just #define needed functionality before and the default is core
functionality only (eg: pgapplication :) which could possibly be disabled
if all one is doing is extending a widget...

at this point a lot of the cross-platform handling is dependant on this
anyways with "paraconfig_gnu.h" defining platform details.  I've added
more than a few #defines here, so take a peek :)

Compiling this lib takes a lot of space already.  Bloating the compile
more by forcing one to include unnecessary objects could... worsen
things... on some poorly-designed (memory/space) wise compilers such as
GNU C++.  (-compiled- binaries are efficient - it's the object
code that isn't).  I'm already running into probs with the 150+ megabytes
needed to compile paragui on my computer *sigh*

so my vote is in favour of adding a single header (say, renaming paragui.h
to paragui_core.h and making paragui.h the big-loader)
BUT - don't force it.  Besides small headers are easier to debug and edit
*grin*
so have paragui.h (with -no- required information so it's an OPTIONAL
header) include all the necessary stuff depending on #define.  That make
sense?

G'day, eh? :)
        - Teunis