Re: [Evolution-hackers] Evolution-2.4 Mac Problems/ Release etc

[Shreyas Sriniavasan <sshreyas novell com>]

On Fri, 2005-09-09 at 21:17 +0800, Not Zed wrote:
> Basically iirc, all of the components are setup in a pretty similar
> manner.   There is a -component or -component-factory.c file which has
> the bonobo module init stuff.  I think ... that just needs to be the
> only file in the bonobo-server's .so file, which links to everything
> which is just in a library with a different name than it is now.  So it
> should actually be very little work.
> 
> Assuming it will actually work, but i guess it should.
> 
Hmm.. So when i actually got down to doing it. This is what i found

so i moved mail-component.c mail-component.h and
mail-component-factory.c to a new
directory /evolution/component/ and was about to write a Makefile which
would build 
a new dynamically loadable module (lets call it
libevolution-mail-component.la)  which can link to the
libevolution-mail.so (a shared library and not a dynamically loadable
module) but thats when "reality bit me"

a simple grep for mail-component.h in mail/ shows a bunch of files
#including the file mail-component.h and ofcourse needless to say you
cant have libevolution-mail.so linking against
libevolution-mail-component.la as that was the original demon we tried
to kill anyway.

So the alternative is to move these files into
libevolution-mail-component.la which would make
them inaccessible to the most plugins. And to evaluate which plugins
were to break would be an other task. Clearly my analysis so far have
been only till the 1st node, i havent bothered looking into what files
include the files which include mail-component.h and so,

The aim of the whole exercise to move all the bonobo interfaces to
libevolution-mail-component.la and all the other stuff to
libevolution-mail.so. Clearly the two seem to be rather intertwined. 

Am I misinterpreting something or "houston we have a problem" ?

Thanks,
Shreyas