Build fixes

[Christian Hergert]

Hi,

I noticed while compiling a fresh checkout of the associated gnome
libraries that the generated code requires UnmanagedFunctionPointer; a
2.0+ feature.

The attached patches alter the compiler to be gmcs.

Should I commit these, or is there another problem that is the real
culprit (such as not generating UnmanagedFunctionPointer)?

-- Christian

[gecko-sharp.patch]

---

Index: configure.in
===================================================================
--- configure.in (revision 138462)
+++ configure.in (working copy)
@@ -22,12 +22,12 @@
 AC_MSG_RESULT($from_cvs)
 
 AC_PATH_PROG(MONO, mono)
-AC_PATH_PROG(MCS, mcs)
+AC_PATH_PROG(MCS, gmcs)
 if test "x$MONO" = "x" ; then
   AC_MSG_ERROR([Can't find "mono" in your PATH])
 fi
 if test "x$MCS" = "x" ; then
-  AC_MSG_ERROR([Can't find "mcs" in your PATH])
+  AC_MSG_ERROR([Can't find "gmcs" in your PATH])
 fi
 AC_SUBST(PATH)
 AC_SUBST(LD_LIBRARY_PATH)
@@ -54,7 +54,7 @@
 
 if test "x$has_mono" = "xtrue"; then
 AC_PATH_PROG(RUNTIME, mono, no)
-AC_PATH_PROG(CSC, mcs, no)
+AC_PATH_PROG(CSC, gmcs, no)
 if test `uname -s` = "Darwin"; then
  LIB_PREFIX=
  LIB_SUFFIX=.dylib
Index: gtkmozembed/Makefile.am
===================================================================
--- gtkmozembed/Makefile.am (revision 138462)
+++ gtkmozembed/Makefile.am (working copy)
@@ -1,4 +1,3 @@
-MCS=mcs
 RAW_API=gtkgecko-api.raw
 API=gtkgecko-api.xml
 METADATA=gtkgecko.metadata
@@ -38,7 +37,7 @@
  cp $(top_srcdir)/gecko-sharp.snk .
 
 gecko-sharp.dll: $(build_sources) generated-stamp gecko-sharp.snk
- $(MCS) -nowarn:169 -unsafe -target:library $(GTK_SHARP_LIBS) \
+ $(CSC) -nowarn:169 -unsafe -target:library $(GTK_SHARP_LIBS) \
  $(build_sources) generated/*.cs -out:gecko-sharp.dll
 
 install-data-local:

---

[gnome-desktop-sharp.patch]

---

Index: configure.in
===================================================================
--- configure.in (revision 138462)
+++ configure.in (working copy)
@@ -98,7 +98,7 @@
  RUNTIME=mono
 fi
 
-AC_PATH_PROG(CSC, mcs, no)
+AC_PATH_PROG(CSC, gmcs, no)
 if test `uname -s` = "Darwin"; then
  LIB_PREFIX=
  LIB_SUFFIX=.dylib

---

[gnome-sharp.patch]

---

Index: configure.in.in
===================================================================
--- configure.in.in (revision 138462)
+++ configure.in.in (working copy)
@@ -110,7 +110,7 @@
  RUNTIME=mono
 fi
 
-AC_PATH_PROG(CSC, mcs, no)
+AC_PATH_PROG(CSC, gmcs, no)
 if test `uname -s` = "Darwin"; then
  LIB_PREFIX=
  LIB_SUFFIX=.dylib

---

_______________________________________________
Gtk-sharp-list maillist  -  Gtk-sharp-list@...
http://lists.ximian.com/mailman/listinfo/gtk-sharp-list

[Christian Hoff]

Christian Hergert wrote:
> Hi,
>
> I noticed while compiling a fresh checkout of the associated gnome
> libraries that the generated code requires UnmanagedFunctionPointer; a
> 2.0+ feature.
>  
It's astonishing how many Gtk# trunk users we have :-) . Neville Gao
already filed a bug report
(https://bugzilla.novell.com/show_bug.cgi?id=521449).

We decided to switch to the .net 2.0 profile for Gtk#/GAPI 3. The
problem is that we didn't bump up the GAPI version yet. Attached to the
bug report is a patch, but it still needs to be reviewed.
> The attached patches alter the compiler to be gmcs.
>  
We would also need to alter configure.in.in to search for GAPI 3 instead
of GAPI 2.
> Should I commit these, or is there another problem that is the real
> culprit (such as not generating UnmanagedFunctionPointer)?
>  
I'm not sure if we need to commit all of these. As Mike points out in
the bug report, nearly everything of gnome-sharp is deprecated or
targeted for deprecation. It's probably best to continue building
obsolete libraries against GAPI 2 as it's wide availability is also an
asset. Once we apply the GAPI 3 patch, you can simply reinstall GAPI 2
and everything will continue to work just fine.

As far as gecko-sharp is concerned, I'm not sure what to do. I never
used it and don't know who is maintaining it, either.

Christian



_______________________________________________
Gtk-sharp-list maillist  -  Gtk-sharp-list@...
http://lists.ximian.com/mailman/listinfo/gtk-sharp-list

[Mike Kestner-4]

On Thu, 2009-07-23 at 11:59 +0200, Christian Hoff wrote:
> Christian Hergert wrote:
> >
> We decided to switch to the .net 2.0 profile for Gtk#/GAPI 3. The
> problem is that we didn't bump up the GAPI version yet. Attached to the
> bug report is a patch, but it still needs to be reviewed.

Please commit the s/gapi2/gapi3 patch.  Also, if you don't mind, please
rename the bootstrap to something like bootstrap-2.90 or you can
resurrect bootstrap-for-the-insane, and also add a big bold text message
to the end of it that this bootstrap is unsupported, unstable, with no
scheduled release date and unless people are completely insane, they
should be using the 2-12 branch instead.  ;-)

> > The attached patches alter the compiler to be gmcs.
> >  
> We would also need to alter configure.in.in to search for GAPI 3 instead
> of GAPI 2.
> > Should I commit these, or is there another problem that is the real
> > culprit (such as not generating UnmanagedFunctionPointer)?
> >  
> I'm not sure if we need to commit all of these. As Mike points out in
> the bug report, nearly everything of gnome-sharp is deprecated or
> targeted for deprecation. It's probably best to continue building
> obsolete libraries against GAPI 2 as it's wide availability is also an
> asset. Once we apply the GAPI 3 patch, you can simply reinstall GAPI 2
> and everything will continue to work just fine.

No, please don't commit any patches to bindings to make them depend on
gtk-sharp trunk yet.  gtk-sharp-2-12-branch is the supported branch.  If
you want to track svn versions, that's the branch to track.

--
Mike Kestner <mkestner@...>

_______________________________________________
Gtk-sharp-list maillist  -  Gtk-sharp-list@...
http://lists.ximian.com/mailman/listinfo/gtk-sharp-list