The combined diff is too large to send as an attachment. Below is the patch to 
Make.inc, placeholders are reinplaced in the post-patch phase.

Attachment: Portfile
Description: Binary data


diff -ur spooles.2.2.orig/Make.inc spooles.2.2/Make.inc
--- spooles.2.2.orig/Make.inc   1999-01-18 15:48:55.000000000 -0600
+++ spooles.2.2/Make.inc        2018-11-17 09:36:12.000000000 -0600
@@ -12,7 +12,7 @@
 # for solaris
 #
 # CC = gcc
-  CC = /usr/lang-4.0/bin/cc
+  CC = @@CC@@
 #
 # for sgi
 #
@@ -28,14 +28,14 @@
 # 
 # OPTLEVEL = 
 # OPTLEVEL = -g -v
-  OPTLEVEL = -O
+# OPTLEVEL = -O
 # OPTLEVEL = -xO5 -v
 # OPTLEVEL = -O3
 # OPTLEVEL = -O4
 # CFLAGS = -Wall -g
 # CFLAGS = -Wall -pg
 # CFLAGS = $(OPTLEVEL) -D_POSIX_C_SOURCE=199506L
-  CFLAGS = $(OPTLEVEL)
+  CFLAGS = @@CFLAGS@@
 # CFLAGS = -Wall $(OPTLEVEL)
 #
 #----------------------------------------
@@ -43,15 +43,15 @@
 #  set any load flags
 #
 #  LDFLAGS = -Wl,+parallel -Wl,+tm,spp2000 # for hp exemplar
-   LDFLAGS =
+   LDFLAGS = @@LDFLAGS@@
 #
 #---------------------------------------------------------------------
 #
 #  set any thread libraries
 #
 # THREAD_LIBS =
-# THREAD_LIBS = -D_REENTRANT=199506L -lpthread
-  THREAD_LIBS = -D_POSIX_C_SOURCE=199506L -lpthread
+  THREAD_LIBS = -D_REENTRANT=199506L -lpthread
+# THREAD_LIBS = -D_POSIX_C_SOURCE=199506L -lpthread
 # THREAD_LIBS = -lpthread
 #
 #---------------------------------------------------------------------
@@ -79,15 +79,15 @@
 #  set the ranlib environment
 #  (if ranlib is not needed, we echo the library name)
 #
-# RANLIB = ranlib
-  RANLIB = echo
+  RANLIB = ranlib
+# RANLIB = echo
 #
 #---------------------------------------------------------------------
 #
 #  set suffix rule *.c --> *.o
 #
-.c.o : 
-       $(PURIFY) $(CC) -c $(CFLAGS) $<
+#.c.o : 
+#      $(PURIFY) $(CC) -c $(CFLAGS) $<
 #
 #---------------------------------------------------------------------
 #
@@ -102,8 +102,8 @@
 #
 #  MPI install library
 #
-# MPI_INSTALL_DIR = 
-  MPI_INSTALL_DIR = /usr/local/mpich-1.0.13
+  MPI_INSTALL_DIR = @@PREFIX@@
+# MPI_INSTALL_DIR = /usr/local/mpich-1.0.13
 #
 #---------------------------------------------------------------------
 #
@@ -115,7 +115,7 @@
 #
 # for solaris
 #
-  MPI_LIB_PATH = -L$(MPI_INSTALL_DIR)/lib/solaris/ch_p4
+  MPI_LIB_PATH = -L$(MPI_INSTALL_DIR)/lib/mpich-mp
 #
 # for hp
 #
@@ -142,6 +142,6 @@
 #  MPI include path
 #
 # MPI_INCLUDE_DIR = 
-  MPI_INCLUDE_DIR = -I$(MPI_INSTALL_DIR)/include
+  MPI_INCLUDE_DIR = -I$(MPI_INSTALL_DIR)/include/mpich-mp
 #
 #---------------------------------------------------------------------


Mark Brethen
mark.bret...@gmail.com



> On Nov 25, 2018, at 2:11 AM, Ryan Schmidt <ryandes...@macports.org> wrote:
> 
> How specifically have you patched it? A diff would be more useful for seeing 
> that.
> 
> I do see the line "CC = $(CC)", which is a no-op: it assigns CC to itself. So 
> that's not useful.
> 
> 
> On Nov 17, 2018, at 09:37, Mark Brethen wrote:
> 
>> I’ve patched the Make.inc file. Does it look correct?
>> <Make.inc>
>> 
>> Mark Brethen
>> mark.bret...@gmail.com
>> 
>> 
>> 
>>> On Nov 16, 2018, at 8:20 PM, Mark Brethen <mark.bret...@gmail.com> wrote:
>>> 
>>> So is this port a good candidate for the mpi port group? If not then I’ll 
>>> just whitelist a compiler if the mpi variant is chosen.
>>> 
>>> 
>>> Mark Brethen
>>> mark.bret...@gmail.com
>>> 
>>> 
>>> 
>>>> On Nov 16, 2018, at 5:08 PM, Mark Brethen <mark.bret...@gmail.com> wrote:
>>>> 
>>>> From the SPOOLES documentation:
>>>> 
>>>> The SPOOLES library operates in serial, multithreaded and MPI 
>>>> environments. The code for these three environments is fairly segregated. 
>>>> The MPI directory contains all source and driver code for MPI programs. 
>>>> The MT directory contains all source and driver code for multithreaded 
>>>> programs. All other directories contain serial code.2 The MPI source code 
>>>> is compiled into a spoolesMPI.a library. The multithreaded source code is 
>>>> compiled into a spoolesMT.a library. The serial code is compiled into a 
>>>> spooles.a library.
>>>> 
>>>> I would like to offer the user the option of MT or MPI. The build phase 
>>>> would look something like this:
>>>> 
>>>> if defined(WITH_MPI)
>>>>            cd ${WRKSRC}/MPI/src ; ${SETENV} ${MAKE_ENV} ${MAKE_CMD} -f 
>>>> makeGlobalLib
>>>>            cd ${WRKSRC}_SHARED/MPI/src; ${SETENV} ${MAKE_ENV} ${MAKE_CMD} 
>>>> -f makeGlobalLib
>>>>            cd ${WRKSRC}_SHARED ; ld -Bshareable -o libspooles.so.1 -x 
>>>> -soname libspooles.so.1 --whole-archive spooles.a 
>>>>    
>>>>    else
>>>>            cd ${WRKSRC}; ${SETENV} ${MAKE_ENV} ${MAKE_CMD} global -f 
>>>> makefile
>>>>            cd ${WRKSRC}/MT/src; ${SETENV} ${MAKE_ENV} ${MAKE_CMD} -f 
>>>> makeGlobalLib
>>>>            cd ${WRKSRC}_SHARED; ${SETENV} ${MAKE_ENV} ${MAKE_CMD} global 
>>>> -f makefile)
>>>>            cd ${WRKSRC}_SHARED/MT/src; ${SETENV} ${MAKE_ENV} ${MAKE_CMD} 
>>>> -f makeGlobalLib
>>>>            cd ${WRKSRC}_SHARED ; ld -Bshareable -o libspooles.so.1 -x 
>>>> -soname libspooles.so.1 --whole-archive spooles.a 
>>>>    endif
>>>> 
>>>> There’s probably an easy way to do this using the MPI portgroup. Can you 
>>>> suggest some ports to look at as an example?
>>>> 
>>>> 
>>>> Mark Brethen
>>>> mark.bret...@gmail.com
>>>> 
>>>> 
>>>> 
>>> 
>> 
> 

Reply via email to