comparison functions to always use POSIX strcasecmp(). For non-POSIX
systems that use other functions (strcmpi or stricmp) use a macro to alias
strcasecmp to those. If we can't find any of them (VMS, plus whatever
UNIX doesn't have them) then define our own version in misc.c.
string into the strcache. As a side-effect, many more structure members and
function arguments can/should be declared const.
As mentioned in the changelog, unfortunately measurement shows that this
change does not yet reduce memory. The problem is with secondary expansion:
because of this we store all the prerequisites in the string cache twice.
First we store the prerequisite string after initial expansion but before
secondary expansion, then we store each individual file after secondary
expansion and expand_deps(). I plan to change expand_deps() to be callable
in either context (eval or snap_deps) then have non-second-expansion
targets call expand_deps() during eval, so that we only need to store that
dependency list once.
A few changes from char* to void* where appropriate, and removing of
unnecessary casts.
Much more work on const-ifying the codebase. This round involves some code
changes to make it correct. NOTE!! There will almost certainly be problems
on the non-POSIX ports that will need to be addressed after the const changes
are finished: they will need to be const-ified properly and there may need to
be some changes to allocate memory, etc. as well.
The next (last?) big push for this, still to come, is const-ifying the
filenames in struct file, struct dep, etc. This will allow us to store file
names in the string cache and finally resolve Savannah bug #15182 (make uses
too much memory), among other advantages.
- Add more warnings.
- Rename variables that mask out-scope vars with the same name.
- Remove all casts of return values from xmalloc, xrealloc, and alloca.
- Remove casts of the first argument to xrealloc.
- Convert all bcopy/bzero/bcmp invocations to use memcp/memmove/memset/memcmp.
I decided this feature was too impacting to make the permanent default
behavior. This set of changes makes the default behavior of make the
old behavior (no second expansion). If you want second expansion, you
must define the .SECONDEXPANSION: special target before the first target
that needs it.
This set of changes ONLY fixes explicit and static pattern rules to work
like this. Implicit rules still have second expansion enabled all the
time: I'll work on that next.
Note that there is still a backward-incompatibility: now to get the old
SysV behavior using $$@ etc. in the prerequisites list you need to set
.SECONDEXPANSION: as well.
check for this and exit with an error.
The closeout.c version from gnulib pulls in too much other stuff, and
gnulib requires an ANSI C 89 compliant compiler, while GNU make (so far)
still wants to work on K&R.
cleanups.
If we find a make error (invalid makefile syntax or something like that)
write back any tokens we have before we exit.
If we have waiting jobs (using -j + -l) set an alarm before we sleep on
the read() system call, so we can wake up to check the load and start
waiting jobs, if there are long-running jobs we would otherwise be
waiting for. Suggested by Grant Taylor.
Taylor. There are two forms of this: first, it was possible to lose
tokens when using -j and -l at the same time, because waiting jobs were
not checked when determining whether any jobs were outstanding. Second,
if you had an exported recursive variable that contained a $(shell ...)
function there is a possibility to lose tokens, since a token was taken
but the child list was not updated until after the shell function was
complete.
To resolve this I introduced a new variable that counted the number of
tokens we have obtained, rather than checking whether there were any
children on the list. I also added some sanity checks to make sure we
weren't writing back too many or not enough tokens. And, the master
make will drain the token pipe before exiting and compare the count of
tokens at the end to what was written there at the beginning.
Also:
* Ensure a bug in the environment (missing "=") doesn't cause make to core.
* Rename the .DEFAULT_TARGET variable to .DEFAULT_GOAL, to match the
terminology in the documentation and other variables like MAKECMDGOALS.
* Add documentation of the .DEFAULT_GOAL special variable.
Still need to document the secondary expansion stuff...
I did this by adding intelligence into the algorithm such that the
second expansion was only actually performed when the prerequisite list
contained at least one "$", so we knew it is actually needed.
Without this we were using up a LOT more memory, since every single
target (even ones never used by make) had their file variables
initialized. This also used a lot more CPU, since we needed to create
and populate a new variable hash table for every target.
There is one issue remaining with this feature: it leaks memory. In
pattern_search() we now initialize the file variables for every pattern
target, which allocates a hash table, etc. However, sometimes we
recursively invoke pattern_search() (for intermediate files) with an
automatic variable (alloca() I believe) as the file. When that function
returns, obviously, the file variable hash memory is lost.
* New function: $(info ...)
* Disallow $(eval ...) to create prereq relationships inside command scripts
(caused core dumps)
* Try to allow more tests to succeed in Windows/DOS by sanitizing CRLF and \
* Various bug fixes and code cleanups (see the ChangeLog entry)
Fix references to MINGW #define constants.
Remove WINDOWS32 ifdef from sub_proc.h.
Only add variables to the command line for recursion once.
New features in run_make_test: #PWD# and #MAKEPATH# replacements.
Test the multi-variable fix in the recursion regression test.
POSIX requires that the value of SHELL in the makefile NOT be exported
to sub-commands. Instead, the value in the environment when make was
invoked should be passed to the environment of sub-commands. Note that
make still uses SHELL to _run_ sub-commands; it just doesn't change the
value of the SHELL variable in the environment of sub-commands.
As an extension to POSIX, if the makefile explicitly exports SHELL then
GNU make _will_ use it in the environment of sub-commands.
This commits a number of changes from Earnie Boyd that allows GNU make
to build for MINGW32 systems. Only missing from this commit are the
changes to configure.in etc.; I'm waiting for Earnie to sign papers for
those new files.
Also not here is any README.mingw32 etc. which would explain how to use
this port.
enable the automake ansi2knr capability.
Right now this doesn't quite build using a K&R compiler because of a
problem with the loadavg test program, but the rest of the code works. I'm
asking the automake list about this problem.
Invent a new macro HAVE_DOS_PATHS and change various instances of:
#if defined(WINDOWS) || defined(__MSDOS__)
to use the new macro instead. This should help make the OS/2 port
cleaner, as well.
Invent a cvs-clean maintainer target that tries to get the workspace
back to the state it was in after a CVS checkout.
New language.
GNU make. Also he provides some other performance fixups after doing
some profiling of make on large makefiles.
Modify the test suite to allow the use of Valgrind to find memory problems.
Incorporate "order-only" prerequisites patch. Wrote a test for it.
The test shows what might be a bug in the code; I need to look at it
more closely (anyway it doesn't behave as I expected). Also I haven't
done the docs yet.
New version of the manual, put into the doc subdir.
Enhancements: $(eval ...) and $(value ...) functions, various bug
fixes, etc. See the ChangeLog.
More to come.
properly.
Fix configure: allow cross-compilation; fix getloadavg (still needs _lots_
of work!)
Let $(call ...) functions to be self-referencing. Lets us do transitive
closures, for example.
Fix an assert() in an obscure use of -q
Handling of double-colon timestamp updates was broken in a bizarre way.
Store arguments to $(call ...) functions in simple variables, not recursive
variables.
* Update maintainers build process; remove GNUmakefile. Require builders to
run automake && autoreconf by hand.
* Use AC_SUBST_FILE to get the maintMakefile included, rather than GNU
make's include directive, which conflicts with automake 1.4's include
directive.
* main.c (print_version): Print out bug-reporting address.
Mon Aug 26 19:55:47 1996 Roland McGrath <roland@baalperazim.frob.com>
* main.c (print_data_base): Don't declare ctime; headers do it for us
already.
* dir.c: WIN32 does not support inode. For now, fully qualified
pathname along with st_mtime will be keys for files.
Fixed problem where vpath can be confused when files
are added to a directory after the directory has already been
read in. The code now attempts to reread the directory if it
discovers that the datestamp on the directory has changed since
it was cached by make. This problem only seems to occur on WIN32
right now so it is lumped under port #ifdef WIN32.
* function.c: WIN32: call subproc library (CreateProcess()) instead of
fork/exec.
* job.c: WIN32: Added the code to do fork/exec/waitpid style processing
on WIN32 systems via calls to subproc library.
* main.c: WIN32: Several things added here. First, there is code
for dealing with PATH and SHELL defaults. Make tries to figure
out if the user has %PATH% set in the environment and sets it to
%Path% if it is not set already. Make also looks to see if sh.exe
is anywhere to be found. Code path through job.c will change
based on existence of a working Bourne shell. The checking for
default shell is done twice: once before makefiles are read in
and again after. Fall back to MSDOS style execution mode if no sh.exe
is found. Also added some debug support that allows user to pause make
with -D switch and attach a debugger. This is especially useful for
debugging recursive calls to make where problems appear only in the
sub-make.
* make.h: WIN32: A few macros and header files for WIN32 support.
* misc.c: WIN32: Added a function end_of_token_w32() to assist
in parsing code in read.c.
* read.c: WIN32: Fixes similar to MSDOS which allow colon to
appear in filenames. Use of colon in filenames would otherwise
confuse make.
* remake.c: WIN32: Added include of io.h to eliminate compiler
warnings. Added some code to default LIBDIR if it is not set
on WIN32.
* variable.c: WIN32: Added support for detecting Path/PATH
and converting them to semicolon separated lists for make's
internal use. New function sync_Path_environment()
which is called in job.c and function.c before creating a new
process. Caller must set Path in environment since we don't
have fork() to do this for us.
* vpath.c: WIN32: Added detection for filenames containing
forward or backward slashes.
* NMakefile: WIN32: Visual C compatible makefile for use with nmake.
Use this to build GNU make the first time on Windows NT or Windows 95.
* README.WIN32: WIN32: Contains some helpful notes.
* build_w32.bat: WIN32: If you don't like nmake, use this the first
time you build GNU make on Windows NT or Windows 95.
* config.h.WIN32: WIN32 version of config.h
* subproc.bat: WIN32: A bat file used to build the
subproc library from the top-level NMakefile. Needed because
WIndows 95 (nmake) doesn't allow you to cd in a make rule.
* w32/include/dirent.h
* w32/compat/dirent.c: WIN32: opendir, readdir, closedir, etc.
* w32/include/pathstuff.h: WIN32: used by files needed functions
defined in pathstuff.c (prototypes).
* w32/include/sub_proc.h: WIN32: prototypes for subproc.lib functions.
* w32/include/w32err.h: WIN32: prototypes for w32err.c.
* w32/pathstuff.c: WIN32: File and Path/Path conversion functions.
* w32/subproc/build.bat: WIN32: build script for subproc library
if you don't wish to use nmake.
* w32/subproc/NMakefile: WIN32: Visual C compatible makefile for use
with nmake. Used to build subproc library.
* w32/subproc/misc.c: WIN32: subproc library support code
* w32/subproc/proc.h: WIN32: subproc library support code
* w32/subproc/sub_proc.c: WIN32: subproc library source code
* w32/subproc/w32err.c: WIN32: subproc library support code
* main.c: (main), variable.c Changed handling of ENV-vars. Make
stores now the names of the variables only and reads their contents
when they are accessed to reflect that these variables are really
global (ie. they CAN change WHILE make runs !) This handling is
made in lookup_variable()
* GNUmakefile (globfiles): Add AmigaDOS support files.
(distfiles): Add $(amigafiles).
(amigafiles): New variable.
Thu Nov 7 10:18:16 1995 Aaron Digulla <digulla@fh-konstanz.de>
* Added Amiga support in commands.c, dir.c, function.c,
job.c, main.c, make.h, read.c, remake.c
* commands.c: Amiga has neither SIGHUP nor SIGQUIT
* dir.c: Amiga has filenames with Upper- and Lowercase,
but "FileName" is the same as "filename". Added strieq()
which is use to compare filenames. This is like streq()
on all other systems. Also there is no such thing as
"." under AmigaDOS.
* function.c: On Amiga, the environment is not passed as envp,
there are no pipes and Amiga can't fork. Use my own function
to create a new child.
* job.c: default_shell is "" (The system automatically chooses
a shell for me). Have to use the same workaround as MSDOS for
running batch commands. Added HAVE_SYS_PARAM_H. NOFILE isn't
known on Amiga. Cloned code to run children from MSDOS. Own
version of sh_chars[] and sh_cmds[]. No dup2() or dup() on Amiga.
* main.c: Force stack to 20000 bytes. Read environment from ENV:
device. On Amiga, exec_command() does return, so I exit()
afterwards.
* make.h: Added strieq() to compare filenames.
* read.c: Amiga needs special extension to have passwd. Only
one include-dir. "Makefile" and "makefile" are the same.
Added "SMakefile". Added special code to handle device names (xxx:)
and "./" in rules.
* remake.c: Only one lib-dir. Amiga link-libs are named "%s.lib"
instead of "lib%s.a".
* main.c, rule.c, variable.c: Avoid floats at all costs.
* vpath.c: Get rid of as many alloca()s as possible.
(decode_switches): Set optind to 0 to reinitialize getopt, not to 1. When
getopt_long returns EOF, break the loop and handle remaining args with a
simple second loop.
Take new arg; if nonzero, also double $s.
(main): Define MAKEOVERRIDES from command_variables here.
(define_makeflags): Don't use command_variables here; instead write a
reference $(MAKEOVERRIDES) in MAKEFLAGS. Make vars recursive.
returns EOF. Initialize C to zero before loop; in loop if C is EOF, set
optarg from ARGV[optind++], else call getopt_long.
(decode_env_switches): Use variable_expand instead of
allocated_variable_expand. Allocate a fresh buffer to copy split words
into; scan characters by hand to break words and debackslashify.
(shell_quote): New function.
(define_makeflags): Allocate doubled space for switch args, and command
variable names and values; use shell_quote to quote those things.
(command_variables): New type and variable.
(decode_switches, decode_env_switches): After making a variable definition,
record the struct variable pointer in the command_variables chain.
(define_makeflags): If ALL, write variable definitions for
command_variables.
(goals, lastgoal): New static variables (moved from auto in main).
(main): Don't process OTHER_ARGS at all.
Don't set variable MAKEOVERRIDES at all; define MAKE to just $(MAKE_COMMAND).
(init_switches): Prepend a - (return in order) instead of a + (require order).
(decode_switches): Don't set OTHER_ARGS at all.
Grok '\0' return from getopt_long as non-option argument; try variable
definition and (if !ENV) enter goal targets here.
(decode_env_switches): Use allocated_variable_expand to store value.
Use find_next_token to simplify word-splitting loop.
Don't prepend a dash to uninterpreted value.
Instead, if split into only one word, try variable definition and failing
that prepend a dash to the word and pass it to decode_switches as a single arg.
