[PATCH] independence from gmake

Adriaan de Groot adridg at cs.kun.nl
Wed Apr 24 19:30:40 BST 2002


On Wednesday 24 April 2002 19:35, Michael Matz wrote:
> We were eager to try it with pmake ;)

Yup, and the BSD folks have been spending all day trying to get things to 
compile again. Of course, this is good for my stats in the stupid 
cvs-percentile-competition :) 

> > am_edit needs some patching as well.
>
> Yep, something I forgot.  I looked at it and most of it seems safe.  Just
> some questions: did you try srcdir!=buildir and srcdir==builddir?  The
> addition of target x for each target $(srcdir)/x has the potential to
> change something in that respect.

Only with build=src. Lauri claims to have used my am_edit with build != src. 
I'd prefer to give this a shot on at least one other system before daring to 
commit is. Look at the amount of grief the automake stuff has given us today 
already. There are better things to do than to correct basic configure and 
make errors.


> The part which adds the IDL files (i.e. _skel.cpp and _stub.h) looks
> wrong.  

Quite possible. That's a version from a day or two ago and it was never meant 
to be "ready for prime time".

> First: you add both these files at once to $othermox (I believe
> you want to collect there all generated files, or?), whereas each

Yes, I'm collecting all the generated sources -- mocs and ui and kidl stuff -- 
into one target. On the one hand this is a remnant from a version of the 
am_edit hacks where all-mocs was built first, but that meant that you had to 
build final -- the recursive call to all-am messed that approach up. On the 
other hand it *seems* like a possibly useful target.

> iteration of that loop only creates one file (and possibly not both are
> created at all).  This is something I'm not entirely sure about in the

Good point. I'll take a look at *why* we're in the loop and add the relevant 
one. 

> moment. But second: You add _skel.cpp, whereas the created file is
> called _skel.$cxxsuffix.  

Fixed in a more recent version. Seriously, the am_edit that went to 
-core-devel was a snapshot of what I had working *right* *then*. Afterwards I 
went on to build more KDE modules, and the list of changes grew as I 
discovered more things that were needed to get it to work with BSD make.


> And why don't you add e.g. the _signals.h files?

Because BSD make never broke on a missing _signals.h file. Since I've never 
used DCOP signals, those kind of files are not foremost in my mind.








More information about the kde-core-devel mailing list