Re: [cp-patches] [patch] refactor text peers to remove GtkTextComponentPeer

[Thomas Fitzsimmons]

On Tue, 2005-07-05 at 07:43 +0200, Mark Wielaard wrote:
> Hi,
> 
> On Mon, 2005-07-04 at 19:57 -0400, Thomas Fitzsimmons wrote:
> > GtkTextComponentPeer is unnecessary since the GTK widgets, GtkTextView
> > and GtkEntry are not related.  This patch refactors the
> > GtkTextComponentPeer code into GtkTextAreaPeer and GtkTextFieldPeer.  I
> > applied this to mainline.
> 
> Please be a bit more careful. This broke the build since you forgot to
> change the include file generation. And --enable-Werror

Should I be building with --enable-Werror?  I plan to commit another
patch today that will eliminate these warnings and I thought it was
acceptable to introduce temporary warnings.  If not, --enable-Werror
should be the default.

>  since
> classpath_gtk_component_connect_signals and the nonfocus variant are not
> actually implemented. Fixed as follows:
> 
> 2005-07-05  Mark Wielaard  <address@hidden>
> 
>        * include/Makefile.am (GTKPEER_H_FILES): Removed GtkTextComponentPeer.
>        * include/gnu_java_awt_peer_gtk_GtkTextComponentPeer.h: Removed.
>        * gnu/java/awt/peer/gtk/GtkComponentPeer.java
>        (classpath_gtk_component_connect_nonfocus_signals): Mark argument
>        unused.
>        (classpath_gtk_component_connect_signals): Likewise.
> 
> When we switch to gcj 4.0+ we can make --enable-regen-headers the
> default with gcjh to catch this kind of thing more easily.

That won't help in this case.  I was building with
--enable-regen-headers.  For some reason, removing
GtkTextComponentPeer.java does not cause
gnu_java_awt_peer_gtk_GtkTextComponentPeer.h to be regenerated.  I
prefer how libgcj handles these headers:

gtk_jni_headers = $(patsubst %.java,jniinclude/%.h,$(subst 
/,_,$(gtk_awt_peer_source_files)))

Anyway, from now on I'll do clean builds before committing, when making
header file changes.

Thanks for fixing,
Tom