summaryrefslogtreecommitdiff
path: root/README.VMS
diff options
context:
space:
mode:
Diffstat (limited to 'README.VMS')
-rw-r--r--README.VMS625
1 files changed, 398 insertions, 227 deletions
diff --git a/README.VMS b/README.VMS
index 2ff553c..bec405d 100644
--- a/README.VMS
+++ b/README.VMS
@@ -1,284 +1,455 @@
-This version of GNU make has been tested on
-OpenVMS V8.3 (Alpha) and V8.4 (Integrity).
+Overview: -*-text-mode-*-
+---------
-Build instructions
-------------------
-Make a 1st version
- $ @makefile.com ! ignore any compiler and/or linker warning
- $ copy make.exe 1st-make.exe
- Use the 1st version to generate a 2nd version
- $ mc sys$disk:[]1st-make clean ! ignore any file not found messages
- $ mc sys$disk:[]1st-make
- Verify your 2nd version
- $ copy make.exe 2nd-make.exe
- $ mc sys$disk:[]2nd-make clean
- $ mc sys$disk:[]2nd-make
- Don't use the HP C V7.2-001 compiler, which has an incompatible change
- how __STDC__ is defined. This results at least in compile time warnings.
-
-Changes since GNU make 3.82
+ This version of GNU make has been tested on:
+ OpenVMS V8.3/V8.4 (Alpha) and V8.4 (Integrity) AND V7.3 (VAX)
+
+ This version of GNU Make is intended to be run from DCL to run
+ make scripts with a special syntax that is described below. It
+ likely will not be able to run unmodified Unix makefiles.
+
+ There is an older implementation of GNU Make that was ported to GNV.
+ Work is now in progress to merge that port to get a single version
+ of GNU Make available. When that merge is done, GNU Make will auto
+ detect that it is running under a Posix shell and then operate as close to
+ GNU Make on Unix as possible.
+
+ The descriptions below are for running GNU make from DCL or equivalent.
+
+
+Recipe differences:
+-------------------
+
+ GNU Make for OpenVMS can not currently run native Unix make files because of
+ differences in the implementation that it is not aware of the GNV packages.
+
+ I am trying to document the current behavior in this section. This is based
+ on the information in the file NEWS. and running the test suite.
+ TODO: More tests are needed to validate and demonstrate the OpenVMS
+ expected behavior.
+
+ The format for recipes are a combination of Unix macros, a subset of
+ simulated UNIX commands, some shell emulation, and OpenVMS commands.
+ This makes the resulting makefiles unique to the OpenVMS port of GNU make.
+
+ If you are creating a OpenVMS specific makefile from scratch, you should also
+ look at MMK (Madgoat Make) available at https://github.com/endlesssoftware/mmk
+ MMK uses full OpenVMS syntax and a persistent subprocess is used for the
+ recipe lines, allowing multiple line rules.
+
+ The default makefile search order is "makefile.vms", "gnumakefile",
+ "makefile". TODO: See if that lookup is case sensitive.
+
+ When Make is invoked from DCL, it will create a foreign command
+ using the name of executable image, with any facility prefix removed,
+ for the duration of the make program, so it can be used internally
+ to recursively run make(). The macro MAKE_COMMAND will be set to
+ this foreign command.
+
+ When make is launched from an exec*() command from a C program,
+ the foreign command is not created. The macro MAKE_COMMAND will be
+ set to the actual command passed as argv[0] to the exec*() function.
+
+ If the DCL symbol or logical name GNV$MAKE_USE_MCR exists, then
+ the macro MAKE_COMMAND will be set to be an "MCR" command with the
+ absolute path used by DCL to launch make. The foreign command
+ will not be created.
+
+ The macro MAKE is set to be the same value as the macro MAKE_COMMAND
+ on all platforms.
+
+ Each recipe command is normally run as a separate spawned processes,
+ except for the cases documented below where a temporary DCL command
+ file may be used.
+
+ BUG: Testing has shown that the commands in the temporary command files
+ are not always created properly. This issue is still under investigation.
+
+ Any macros marked as exported are temporarily created as DCL symbols
+ for child images to use. DCL symbol substitution is not done with these
+ commands.
+ TODO: Add symbol substitution.
+
+ When a temporary DCL command file is used, DCL symbol substitution
+ will work.
+
+ Command lines of excessive length are broken and written to a command file
+ in sys$scratch:. There's no limit to the lengths of commands (and no need
+ for .opt files :-) any more.
+
+ The '<', '>' and '>>' redirection has been implemented by using
+ temporary command files. These will be described later.
+
+ The DCL symbol or logical name GNV$MAKE_USE_CMD_FILE when set to a
+ string starting with one of '1','T', or 'E' for "1", "TRUE", or "ENABLE",
+ then temporary DCL command files are always used for running commands.
+ In this case, the exported environment environment variables are
+ created by command file. BUG: Environment variables that hold values
+ with dollar signs in them are not exported correctly.
+
+ GNU Make generally does text comparisons for the targets and sources. The
+ make program itself can handle either Unix or OpenVMS format filenames, but
+ normally does not do any conversions from one format to another.
+ TODO: The OpenVMS format syntax handling is incomplete.
+ TODO: ODS-5 EFS support is missing.
+ BUG: The internal routines to convert filenames to and from OpenVMS format
+ do not work correctly.
+
+ Note: In the examples below, line continuations such as a backslash may have
+ been added to make the examples easier to read in this format.
+ BUG: That feature does not completely work at this time.
+
+ Since the OpenVMS utilities generally expect OpenVMS format paths, you will
+ usually have to use OpenVMS format paths for rules and targets.
+ BUG: Relative OpenVMS paths may not work in targets, especially combined
+ with vpaths. This is because GNU make will just concatenate the directories
+ as it does on Unix.
+
+ The variables $^ and $@ separate files with commas instead of spaces.
+ While this may seem the natural thing to do with OpenVMS, it actually
+ causes problems when trying to use other make functions that expect the
+ files to be separated by spaces. If you run into this, you need the
+ following workaround to convert the output.
+ TODO: Look at have the $^ and $@ use spaces like on Unix and have
+ and easy to use function to do the conversions and have the built
+ in OpenVMS specific recipes and macros use it.
+
+ Example:
+
+comma := ,
+empty :=
+space := $(empty) $(empty)
+
+foo: $(addsuffix .3,$(subs $(comma),$(space),$^)
+
+
+ Makefile variables are looked up in the current environment. You can set
+ symbols or logicals in DCL and evaluate them in the Makefile via
+ $(<name-of-symbol-or-logical>). Variables defined in the Makefile
+ override OpenVMS symbols/logicals.
+
+ OpenVMS logical and symbols names show up as "environment" using the
+ origin function. when the "-e" option is specified, the origion function
+ shows them as "environment override". On Posix the test scripts indicate
+ that they should show up just as "environment".
+
+ When GNU make reads in a symbol or logical name into the environment, it
+ converts any dollar signs found to double dollar signs for convenience in
+ using DCL symbols and logical names in recipes. When GNU make exports a
+ DCL symbol for a child process, if the first dollar sign found is followed
+ by second dollar sign, then all double dollar signs will be convirted to
+ single dollar signs.
+
+ The variable $(ARCH) is predefined as IA64, ALPHA or VAX respectively.
+ Makefiles for different OpenVMS systems can now be written by checking
+ $(ARCH). Since IA64 and ALPHA are similar, usually just a check for
+ VAX or not VAX is sufficient.
+
+ You may have to update makefiles that assume VAX if not ALPHA.
+
+ifeq ($(ARCH),VAX)
+ $(ECHO) "On the VAX"
+else
+ $(ECHO) "On the ALPHA or IA64"
+endif
+
+ Empty commands are handled correctly and don't end in a new DCL process.
+
+ The exit command needs to have OpenVMS exit codes. To pass a Posix code
+ back to the make script, you need to encode it by multiplying it by 8
+ and then adding %x1035a002 for a failure code and %x1035a001 for a
+ success. Make will interpret any posix code other than 0 as a failure.
+ TODO: Add an option have simulate Posix exit commands in recipes.
+
+ Lexical functions can be used in pipes to simulate shell file test rules.
+
+ Example:
+
+ Posix:
+b : c ; [ -f $@ ] || echo >> $@
+
+ OpenVMS:
+b : c ; if f$$search("$@") then pipe open/append xx $@ ; write xx "" ; close xx
+
+
+ You can also use pipes and turning messages off to silently test for a
+ failure.
+
+x = %x1035a00a
+
+%.b : %.c
+<tab>pipe set mess/nofac/noiden/nosev/notext ; type $^/output=$@ || exit $(x)
+
+
+Runtime issues:
+
+ The OpenVMS C Runtime has a convention for encoding a Posix exit status into
+ to OpenVMS exit codes. These status codes will have the hex value of
+ 0x35a000. OpenVMS exit code may also have a hex value of %x10000000 set on
+ them. This is a flag to tell DCL not to write out the exit code.
+
+ To convert an OpenVMS encoded Posix exit status code to the original code
+ You subtract %x35a000 and any flags from the OpenVMS code and divide it by 8.
+
+ WARNING: Backward-incompatibility!
+ The make program exit now returns the same encoded Posix exit code as on
+ Unix. Previous versions returned the OpenVMS exit status code if that is what
+ caused the recipe to fail.
+ TODO: Provide a way for scripts calling make to obtain that OpenVMS status
+ code.
+
+ Make internally has two error codes, MAKE_FAILURE and MAKE_TROUBLE. These
+ will have the error "-E-" severity set on exit.
+
+ MAKE_TROUBLE is returned only if the option "-q" or "--question" is used and
+ has a Posix value of 1 and an OpenVMS status of %x1035a00a.
+
+ MAKE_FAILURE has a Posix value of 2 and an OpenVMS status of %x1035a012.
+
+ Output from GNU make may have single quotes around some values where on
+ other platforms it does not. Also output that would be in double quotes
+ on some platforms may show up as single quotes on VMS.
+
+ There may be extra blank lines in the output on VMS.
+ https://savannah.gnu.org/bugs/?func=detailitem&item_id=41760
+
+ There may be a "Waiting for unfinished jobs..." show up in the output.
+
+ Error messages generated by Make or Unix utilities may slightly vary from
+ Posix platforms. Typically the case may be different.
+
+ When make deletes files, on posix platforms it writes out 'rm' and the list
+ of files. On VMS, only the files are writen out, one per line.
+ TODO: VMS
+
+ There may be extra leading white space or additional or missing whitespace
+ in the output of recipes.
+
+ GNU Make uses sys$scratch: for the tempfiles that it creates.
+
+ The OpenVMS CRTL library maps /tmp to sys$scratch if the TMP: logical name
+ does not exist. As the CRTL may use both sys$scratch: and /tmp internally,
+ if you define the TMP logical name to be different than SYS$SCRATCH:,
+ you may end up with only some temporary files in TMP: and some in SYS$SCRATCH:
+
+ The default include directory for including other makefiles is
+ SYS$SYSROOT:[SYSLIB] (I don't remember why I didn't just use
+ SYS$LIBRARY: instead; maybe it wouldn't work that way).
+ TODO: A better default may be desired.
+
+ If the device for a file in a recipe does not exist, on OpenVMS an error
+ message of "stat: <file>: no such device or address" will be output.
+
+ Make ignores success, informational, or warning errors (-S-, -I-, or
+ -W-). But it will stop on -E- and -F- errors. (unless you do something
+ to override this in your makefile, or whatever).
+
+
+Unix compatibilty features:
---------------------------
-Fix build problems.
-
-The new feature "Loadable objects" is not yet supported. If you need it,
-please send a change request or submit a bug report.
-
-The new option --output-sync (-O) is accepted but has no effect: GNU make
-for VMS does not support running multiple commands simultaneously.
-
-Changes for GNU make 3.82
-
-Michael Gehre (at VISTEC-SEMI dot COM) supplied a fix for a problem with
-timestamps of object modules in OLBs. The timestamps were not correctly
-adjusted to GMT based time, if the local VMS time was using a daylight saving
-algorithm and if daylight saving was switched off.
-
-John Eisenbraun (at HP dot COM) supplied fixes and and an enhancement to append
-output redirection in action lines.
-
-Rework of ctrl+c and ctrl+y handling.
-
-Fix a problem with cached strings, which showed on case-insensitive file
-systems.
-
-Build fixes for const-ified code in VMS specific sources.
-
-A note on appending the redirected output. With this change, a simple mechanism
-is implemented to make ">>" work in action lines. In VMS there is no simple
-feature like ">>" to have DCL command or program output redirected and appended
-to a file. GNU make for VMS already implements the redirection of output. If
-such a redirection is detected, an ">" on the action line, GNU make creates a
-DCL command procedure to execute the action and to redirect its output. Based
-on that, now ">>" is also recognized and a similar but different command
-procedure is created to implement the append. The main idea here is to create a
-temporary file which collects the output and which is appended to the wanted
-output file. Then the temporary file is deleted. This is all done in the
-command procedure to keep changes in make small and simple. This obviously has
-some limitations but it seems good enough compared with the current ">"
-implementation. (And in my opinion, redirection is not really what GNU make has
-to do.) With this approach, it may happen that the temporary file is not yet
-appended and is left in SYS$SCRATCH. The temporary file names look like
-"CMDxxxxx.". Any time the created command procedure can not complete, this
-happens. Pressing Ctrl+Y to abort make is one case. In case of Ctrl+Y the
-associated command procedure is left in SYS$SCRATCH as well. Its name is
-CMDxxxxx.COM.
-
-Change in the Ctrl+Y handling. The CtrlY handler now uses $delprc to delete all
-children. This way also actions with DCL commands will be stopped. As before
-the CtrlY handler then sends SIGQUIT to itself, which is handled in common
-code.
-
-Change in deleteing temporary command files. Temporary command files are now
-deleted in the vms child termination handler. That deletes them even if
-a Ctrl+C was pressed.
-
-The behavior of pressing Ctrl+C is not changed. It still has only an effect,
-after the current action is terminated. If that doesn't happen or takes too
-long, Ctrl+Y should be used instead.
-
-Changes for GNU make 3.80
-
-. In default.c define variable ARCH as IA64 for VMS on Itanium systems.
-
-. In makefile.vms avoid name collision for glob and globfree.
-
-This is the VMS port of GNU Make done by Hartmut.Becker@compaq.com.
-
-It is based on the specific version 3.77k and on 3.78.1. 3.77k was done
-by Klaus Kämpf <kkaempf@rmi.de>, the code was based on the VMS port of
-GNU Make 3.60 by Mike Moretti.
-
-It was ported on OpenVMS/Alpha V7.1, DECC V5.7-006. It was re-build and
-tested on OpenVMS/Alpha V7.2, OpenVMS/VAX 7.1 and 5.5-2. Different
-versions of DECC were used. VAXC was tried: it fails; but it doesn't
-seem worth to get it working. There are still some PTRMISMATCH warnings
-during the compile. Although perl is working on VMS the test scripts
-don't work. The function $shell is still missing.
-
-There is a known bug in some of the VMS CRTLs. It is in the shipped
-versions of VMS V7.2 and V7.2-1 and in the currently (October 1999)
-available ECOs for VMS V7.1 and newer versions. It is fixed in versions
-shipped with newer VMS versions and all ECO kits after October 1999. It
-only shows up during the daylight saving time period (DST): stat()
-returns a modification time 1 hour ahead. This results in GNU make
-warning messages. For a just created source you will see:
+ The variable $(CD) is implemented as a built in Change Directory
+ command. This invokes the 'builtin_cd' Executing a 'set default'
+ recipe doesn't do the trick, since it only affects the subprocess
+ spawned for that command.
+ TODO: Need more info on how to use and side effects
- $ gmake x.exe
- gmake.exe;1: *** Warning: File 'x.c' has modification time in the future (940582863 > 940579269)
- cc /obj=x.obj x.c
- link x.obj /exe=x.exe
- gmake.exe;1: *** Warning: Clock skew detected. Your build may be incomplete.
+ Unix shell style I/O redirection is supported. You can now write lines like:
+ "<tab>mcr sys$disk:[]program.exe < input.txt > output.txt &> error.txt"
+ BUG: This support is not handling built in make macros with "<" in them
+ properly.
+ Posix shells have ":" as a null command. OpenVMS generates a DCL warning
+ when this is encountered. It would probably be simpler to have OpenVMS just
+ handle this instead of changing all the tests that use this feature.
+ https://savannah.gnu.org/bugs/index.php?41761
-New in 3.78.1:
+ A note on appending the redirected output. A simple mechanism is
+ implemented to make ">>" work in action lines. In OpenVMS there is no simple
+ feature like ">>" to have DCL command or program output redirected and
+ appended to a file. GNU make for OpenVMS already implements the redirection
+ of output. If such a redirection is detected, an ">" on the action line,
+ GNU make creates a DCL command procedure to execute the action and to
+ redirect its output. Based on that, now ">>" is also recognized and a
+ similar but different command procedure is created to implement the
+ append. The main idea here is to create a temporary file which collects
+ the output and which is appended to the wanted output file. Then the
+ temporary file is deleted. This is all done in the command procedure to
+ keep changes in make small and simple. This obviously has some limitations
+ but it seems good enough compared with the current ">" implementation.
+ (And in my opinion, redirection is not really what GNU make has to do.)
+ With this approach, it may happen that the temporary file is not yet
+ appended and is left in SYS$SCRATCH.
-Fix a problem with automatically remaking makefiles. GNU make uses an
-execve to restart itself after a successful remake of the makefile. On
-UNIX systems execve replaces the running program with a new one and
-resets all signal handling to the default. On VMS execve creates a child
-process, signal and exit handlers of the parent are still active, and,
-unfortunately, corrupt the exit code from the child. Fix in job.c:
-ignore SIGCHLD.
+ The temporary file names look like "CMDxxxxx.". Any time the created
+ command procedure can not complete, this happens. Pressing Ctrl+Y to
+ abort make is one case. In case of Ctrl+Y the associated command
+ procedure is left in SYS$SCRATCH as well. Its name is CMDxxxxx.COM.
-Added some switches to reflect latest features of DECC. Modifications in
-makefile.vms.
+ The CtrlY handler now uses $delprc to delete all children. This way also
+ actions with DCL commands will be stopped. As before the CtrlY handler
+ then sends SIGQUIT to itself, which is handled in common code.
-Set some definitions to reflect latest features of DECC. Modifications in
-config.h-vms (which is copied to config.h).
+ Temporary command files are now deleted in the OpenVMS child termination
+ handler. That deletes them even if a Ctrl+C was pressed.
+ TODO: Does the previous section about >> leaving files still apply?
-Added extern strcmpi declaration to avoid 'implicitly declared' messages.
-Modification in make.h.
+ The behavior of pressing Ctrl+C is not changed. It still has only an effect,
+ after the current action is terminated. If that doesn't happen or takes too
+ long, Ctrl+Y should be used instead.
-Default rule for C++, conditionals for gcc (GCC_IS_NATIVE) or DEC/Digital/
-Compaq c/c++ compilers. Modifications in default.c.
-Usage of opendir() and friends, suppress file version. Modifications in dir.c.
+Build Options:
-Added VMS specific code to handle ctrl+c and ctrl+y to abort make.
-Modifications in job.c.
+ Added support to have case sensitive targets and dependencies but to
+ still use case blind file names. This is especially useful for Java
+ makefiles on VMS:
-Added support to have case sensitive targets and dependencies but to
-still use case blind file names. This is especially useful for Java
-makefiles on VMS:
+<TAB>.SUFFIXES :
+<TAB>.SUFFIXES : .class .java
+<TAB>.java.class :
+<TAB><TAB>javac "$<"
+<TAB>HelloWorld.class : HelloWorld.java
- .SUFFIXES :
- .SUFFIXES : .class .java
- .java.class :
- javac "$<
- HelloWorld.class : HelloWorld.java
+ A new macro WANT_CASE_SENSITIVE_TARGETS in config.h-vms was introduced.
+ It needs to be enabled to get this feature; default is disabled.
+ TODO: This should be a run-time setting based on if the process
+ has been set to case sensitive.
-A new macro WANT_CASE_SENSITIVE_TARGETS in config.h-vms was introduced.
-It needs to be enabled to get this feature; default is disabled. The
-macro HAVE_CASE_INSENSITIVE_FS must not be touched: it is still enabled.
-Modifications in file.c and config.h-vms.
-Bootstrap make to start building make is still makefile.com, but make
-needs to be re-made with a make to make a correct version: ignore all
-possible warnings, delete all objects, rename make.exe to a different
-name and run it.
+Unimplemented functionality:
-Made some minor modifications to the bootstrap build makefile.com.
-
-This is the VMS port of GNU Make.
+ The new feature "Loadable objects" is not yet supported. If you need it,
+ please send a change request or submit a bug report.
-It is based on the VMS port of GNU Make 3.60 by Mike Moretti.
+ The new option --output-sync (-O) is accepted but has no effect: GNU make
+ for OpenVMS does not support running multiple commands simultaneously.
-This port was done by Klaus Kämpf <kkaempf@rmi.de>
-There is first-level support available from proGIS Software, Germany.
-Visit their web-site at http://www.progis.de to get information
-about other vms software and forthcoming updates to gnu make.
+Self test failures and todos:
+-----------------------------
-New for 3.77:
+ GNU make was not currently translating the OpenVMS encoded POSIX values
+ returned to it back to the Posix values. I have temporarily modified the
+ Perl test script to compensate for it. This should be being handled
+ internally to Make.
+ TODO: Verify and update the Perl test script.
-/bin/sh style I/O redirection is supported. You can now write lines like
- mcr sys$disk:[]program.exe < input.txt > output.txt &> error.txt
+ The features/parallelism test was failing. OpenVMS is executing the rules
+ in sequence not in parallel as this feature was not implemented.
+ GNU Make on VMS no longer claims it is implemented.
+ TODO: Implement it.
-Makefile variables are looked up in the current environment. You can set
-symbols or logicals in DCL and evaluate them in the Makefile via
-$(<name-of-symbol-or-logical>). Variables defined in the Makefile
-override VMS symbols/logicals !
+ The vpath feature may need the targets to be in OpenVMS format. To be
+ consistent with other target processing, this restriction should be removed.
+ TODO: Verify this after recent changes.
-Functions for file names are working now. See the GNU Make manual for
-$(dir ...) and $(wildcard ...). Unix-style and VMS-style names are
-supported as arguments.
+ The features/vpathgpath test is failing. Reason has not yet been determined.
-The default rules are set up for GNU C. Building an executable from a
-single source file is as easy as 'make file.exe'.
+ The misc/bs-nl test is failing. This is where a line is continued with a
+ backslash.
-The variable $(ARCH) is predefined as ALPHA or VAX resp. Makefiles for
-different VMS systems can now be written by checking $(ARCH) as in
- ifeq ($(ARCH),ALPHA)
- $(ECHO) "On the Alpha"
- else
- $(ECHO) "On the VAX"
- endif
+ The options/dash-e test is failing. Need to determine how to do overrides
+ on VMS.
-Command lines of excessive length are correctly broken and written to a
-batch file in sys$scratch for later execution. There's no limit to the
-lengths of commands (and no need for .opt files :-) any more.
+ The options/dash-k test is failing. Test is not stopping when it should.
-Empty commands are handled correctly and don't end in a new DCL process.
+ The options/dash-n test is failing. The "+" handling is not working.
+ MAKEFLAG appears not to work.
+ Symlink support is not present. Symlinks are supported by OpenVMS 8.3 and
+ later.
-New for 3.76:
+ The targets/INTERMEDIATE and targets/SECONDARY tests are failing.
+ When make deletes files, on posix platforms it writes out 'rm' and the list
+ of files. On vms, only the files are writen out, one per line.
-John W. Eaton has updated the VMS port to support libraries and VPATH.
+ The variables/GNUMAKEFLAGS and variables/MAKE_RESTARTS are failing.
+ The variables/MAKEFILES test is failing. Reason not yet determined.
-To build Make, simply type @makefile. This should compile all the
-necessary files and link Make. There is also a file called
-makefile.vms. If you already have GNU Make built you can just use
-Make with this makefile to rebuild.
+ The variables/MAKEFLAGS test is failing. Looks like the child is failing.
-Here are some notes about GNU Make for VMS:
+ The variables/automatic test is failing.
+ The $^D, $^F, $+D, $+F cases are failing.
-The cd command is supported if it's called as $(CD). This invokes
-the 'builtin_cd' command which changes the directory.
-Calling 'set def' doesn't do the trick, since a sub-shell is
-spawned for this command, the directory is changed *in this sub-shell*
-and the sub-shell ends.
+ The variables/undefine test is failing. Undefine of multi-line define fails.
-Libraries are not supported. They were in GNU Make 3.60 but somehow I
-didn't care porting the code. If there is enough interest, I'll do it at
-some later time.
+ Error messages should be supressed with the "-" at the beginning of a line.
+ On openVMS they were showing up. TODO: Is this still an issue?
-The variable $^ separates files with commas instead of spaces (It's the
-natural thing to do for VMS).
+ The internal vmsify and unixify OpenVMS to/from UNIX are not handling logical
+ names correctly.
-See defaults.c for VMS default suffixes and my definitions for default
-rules and variables.
-The shell function is not implemented yet.
+Build instructions:
+------------------
+
+ Don't use the HP C V7.2-001 compiler, which has an incompatible change
+ how __STDC__ is defined. This results at least in compile time warnings.
-Load average routines haven't been implemented for VMS yet.
+Make a 1st version
+ $ @makefile.com ! ignore any compiler and/or linker warning
+ $ copy make.exe 1st-make.exe
-The default include directory for including other makefiles is
-SYS$SYSROOT:[SYSLIB] (I don't remember why I didn't just use
-SYS$LIBRARY: instead; maybe it wouldn't work that way).
+ Use the 1st version to generate a 2nd version as a test.
+ $ mc sys$disk:[]1st-make clean ! ignore any file not found messages
+ $ mc sys$disk:[]1st-make
+
+ Verify your 2nd version by building Make again.
+ $ copy make.exe 2nd-make.exe
+ $ mc sys$disk:[]2nd-make clean
+ $ mc sys$disk:[]2nd-make
+
+
+Running the tests:
+------------------
-The default makefiles make looks for are: makefile.vms, gnumakefile,
-makefile., and gnumakefile. .
+ Running the tests on OpenVMS requires the following software to be installed
+ as most of the tests are Unix oriented.
-The stat() function and handling of time stamps in VMS is broken, so I
-replaced it with a hack in vmsfunctions.c. I will provide a full rewrite
-somewhere in the future. Be warned, the time resolution inside make is
-less than what vms provides. This might be a problem on the faster Alphas.
+ * Perl 5.18 or later.
+ https://sourceforge.net/projects/vmsperlkit/files/
+ * GNV 2.1.3 + Updates including a minimum of:
+ * Bash 4.3.30
+ * ld_tools 3.0.2
+ * coreutils 8.21
+ https://sourceforge.net/p/gnv/wiki/InstallingGNVPackages/
+ https://sourceforge.net/projects/gnv/files/
-You can use a : in a filename only if you precede it with a backslash ('\').
-E.g.- hobbes\:[bogas.files]
+ As the test scripts need to create some foreign commands that persist
+ after the test is run, it is recommend that either you use a subprocess or
+ a dedicated login to run the tests.
-Make ignores success, informational, or warning errors (-S-, -I-, or
--W-). But it will stop on -E- and -F- errors. (unless you do something
-to override this in your makefile, or whatever).
+ To get detailed information for running the tests:
-Remote stuff isn't implemented yet.
+ $ set default [.tests]
+ $ @run_make_tests help
-Multiple line DCL commands, such as "if" statements, must be put inside
-command files. You can run a command file by using \@.
+ Running the script with no parameters will run all the tests.
+ After the the test script has been run once in a session, assuming
+ that you built make in sys$disk:[make], you can redefined the
+ "bin" logical name as follows:
-VMS changes made for 3.74.3
+ $ define bin sys$disk:[make],gnv$gnu:[bin]
-Lots of default settings are adapted for VMS. See default.c.
+ Then you can use Perl to run the scripts.
-Long command lines are now converted to command files.
+ $ perl run_make_tests.pl
-Comma (',') as a separator is now allowed. See makefile.vms for an example.
--------------------------------------------------------------------------------
-Copyright (C) 1996-2014 Free Software Foundation, Inc.
-This file is part of GNU Make.
+Acknowlegements:
+----------------
-GNU Make is free software; you can redistribute it and/or modify it under the
-terms of the GNU General Public License as published by the Free Software
-Foundation; either version 3 of the License, or (at your option) any later
-version.
+See NEWS. for details of past changes.
-GNU Make is distributed in the hope that it will be useful, but WITHOUT ANY
-WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
-A PARTICULAR PURPOSE. See the GNU General Public License for more details.
+ These are the currently known contributers to this port.
-You should have received a copy of the GNU General Public License along with
-this program. If not, see <http://www.gnu.org/licenses/>.
+ Hartmut Becker
+ John Malmberg
+ Michael Gehre
+ John Eisenbraun
+ Klaus Kaempf
+ Mike Moretti
+ John W. Eaton