Hello, as you may know, I have been working on the auto-dependency generation
for large build system using GNU make, along the lines of the relevant section
of the GNU make manual, as well as the excellent survey article
http://make.paulandlesley.org/autodep.html.
Well, I have finished but sud
Hello,
I am working on a project which has more than 100M lines of code, more than
100K
files and, more than 1000 makefiles. Makefiles apparently have bugs (may build
too little or too much), are slow, messy, and violate "Paul's rules" and common
decency on a routine basis. I am asked to "do
- Original Message
From: David Boyce
> If one of your platforms is Linux I may have a good starting place for
you. I recently released an open source tool which does basically what
Thank you all for helpful hints, I am looking at the autoconf/automake solution
suggested earlier and then
Hello,
we are using GNU Make 3.82, and I just asked the admin to reinstall it, just in
case the installation went south. But still, I am getting a segmentation
fault.
I will show you below a smallest possible sample case, but, the fault happens a
lot in similar cases. So I am hoping someon
I forgot to add,
the file sw/foobar.cpp
is already present (empty file for this example)
___
Help-make mailing list
Help-make@gnu.org
https://lists.gnu.org/mailman/listinfo/help-make
--- On Thu, 2/7/13, Michael Stahl wrote:
> if the build system has generated a file, then it should be
> able to
> re-generate it if needed; that's its job.
Well, let's agree to disagree here, then. If you delete a crucial intermediate
file without knowing what you are doing, then in my op
--- On Fri, 2/8/13, Michael Stahl wrote:
> > [mark@grid003 ~]$ cat Makefile
> > bar1: bar
> > touch $@
> >
> > bar: foo
> > foo : baz
> > touch foo bar
> > [mark@grid003 ~]$ touch bar1 foo bar
> > [mark@grid003 ~]$ touch baz
> > [mark@grid003 ~]$ make bar1
> > touch foo bar
> > touch bar1
>
>
--- On Fri, 2/8/13, Gerhard Fiedler wrote:
> Having to rely on "clean" builds when incremental builds
> could do the
> job just fine is IMO either an artifact of an insufficient
> build system
> or a result of a bug (or missing feature) in make.
Gerhard, IMHO you are misinterpreting me. I am
Yes Michael, the exact solution I quoted, has a problem in the case you showed.
Thank you for pointing this out and sorry for dragging this obvious case.
But then of course, that's not what I am actually using :) I forgot. I do
this, and I believe that is the correct solution to the "gapin
Thank you Gerhard, this is great that we have it clarified. So you want the
users to be able to delete intermediate files, and I tell them " that is not
supported". In your case, indeed GNU Make seems to be deficient. For me it's
fine, although, Eric has a nice feature there, no question. In
rs without dup rules
$(sort $(MAKE_DIRS)):
mkdir $@
- Original Message ----
From: Mark Galeck (CW)
To: "help-make@gnu.org"
Well, OK, I tried this, but does not work deeper than 1 level of directories, I
guess because the chain of implicit rules does n
Yes but it is not going to be very easy.
There is a principle of good design in software engineering, that each piece of
information should be defined in one place only, and then other users of that
information get it from there.
You want to define the name of your "out file" in your C code
>make -f Makefile-unittests test.
(...)
>make -f Makefile
(...)
>a) The application build fails when the build of the tests fails.
call Makefile-unittests recursively from Makefile, to do this:
make a rule like this in Makefile:
.PHONY: test
test:
$(MAKE) -f Makefile-unittests $@
But: I am as
>
> Assigning to MAKEFLAGS only affects sub-makes, despite this (wrong)
> paragraph from the documentation:
>
> MAKEFLAGS
> The flags given to make. You can set this in the environment
> or a makefile to set flags. See Communicating Options to a
I disagree that this paragraph is wrong.
The target-specific variables are only in effect in recipes, not in
prerequisites and other targets, so the above won't work.
You have to hack your own "target-specific" variable name, depending on the
target, like this:
mwb_DDI_SUFFIX := _mwb
$(foreach goal, owb mwb, $(eval $(goal):
../../rootf
Hello,
as root, I replaced the default /bin/sh with a shell compiled by me with custom
changes. I expected GNU Make will use that shell by default , but it does not
- it still uses the original /bin/sh (which was really bash) even though it is
not there anymore.
Why, and how to use a different
, and yes, it is /bin/bash on my machine too.
From: Tim Murphy
To: Mark Galeck
Cc: "help-make@gnu.org"
Sent: Tuesday, October 13, 2015 2:55 AM
Subject: Re: how to use a different /bin/sh with GNU Make?
/bin/sh is a link to /bin/bash on my machine
if you type echo $SH
I did some more experiments and this is bizarre...
(to remind, I am not changing Make SHELL variable value, I am just manipulating
the contents of /bin/sh hoping to get different behaviour)
calling Make function $(shell) yes that does use the modified shellat the same
time, recipe line invocation
pdated, so that meant, the
current modified shell, was not interpreting any command.
From: Paul Smith
To: Mark Galeck
Cc: "help-make@gnu.org"
Sent: Tuesday, October 13, 2015 5:02 AM
Subject: Re: how to use a different /bin/sh with GNU Make?
On Tue, 2015-10-13 at 04:3
f I can help it.
Is there another way to get massively recursive GNU Make system to accept a
custom shell.
Thank you,
Mark
From: Paul Smith
To: Mark Galeck
Cc: "help-make@gnu.org"
Sent: Tuesday, October 13, 2015 5:02 AM
Subject: Re: how to use a different /bin/sh with GNU
>If you want to force make to always start a shell and never use the fast
path the best way is to name your shell something else, like /bin/mgsh
or something, then set make's SHELL variable to that path. Since make
is no longer sure if your shell is a POSIX shell, it will never use the
fast path
>I assume this means that you can set it for the top-level make if
there is only one top-level make,
No I can't, the $(SHELL) variable is special and is not passed in the usual
way. You have to set it separately each time, or pass it every time on $(MAKE)
recipe line.
awned. So please forgive me for believing in the
literal word of the manual.
Again, thank you for pointing me how to hack this thing. I really appreciate.
Mark
From: Paul Smith
To: Mark Galeck
Cc: "help-make@gnu.org"
Sent: Tuesday, October 13, 2015 5:42 AM
Subject:
Hello, I have another question today...
I change source of GNU Make a little (thank you Paul for the suggestion!).
I run ./configure then make (existing make) to build the changed sources. But
then
>./make
immediately segfaults...
Yes this is some non-standard Linux system - a chroot jail in
> Does the resulting make pass its regression
tests ("make check" after you build it)? If not, then likely there's
some kind of environmental problem on your system. If it does but fails
on your makefiles, then likely there's a bug in 4.1 which is tickled by
your makefiles.
No any make invocat
egfaulting, means the *nix system is messed up (despite
that configure succeeds), then I will just use 3.81 there, it does compile and
run.
I don't really need to fix that *nix system, it is just a temporary thing.
Thank you
From: Tim Murphy
To: Paul D. Smith
Cc: Mark Galeck ; &qu
Subject: Re: how to use a different /bin/sh with GNU Make?
On Tue Oct 13 12:16:48 2015, mark_gal...@pacbell.net (Mark Galeck) wrote:
> I mean, I guess my big problem is this.
> I have a massively GNU Make recursive system. (Yes I know it is a bad thing;
> I did not write it myself).
&g
But I can't spend
time on it.
From: Tim Murphy
To: Mark Galeck
Cc: Paul D. Smith ; "help-make@gnu.org"
Sent: Tuesday, October 13, 2015 12:03 PM
Subject: Re: ./configure Make sources runs, but compiled make segfaults - is
this a bug?
Well, it could be a bug wit
yes somehow the simple assignment SHELL= on the top-level
make comamnd line, does propagate.
Thank you!
From: David Boyce
To: Mark Galeck
Cc: Reinier Post ; "help-make@gnu.org"
Sent: Tuesday, October 13, 2015 12:23 PM
Subject: Re: how to use a different /bin/sh with GNU Mak
g for 3.81, it did not work. I can't use 4.1 because it segfaults
for me.
Have to understand 3.81 code better...
Thank you everybody for your help
From: David Boyce
To: Mark Galeck
Cc: Reinier Post ; "help-make@gnu.org"
Sent: Tuesday, October 13, 2015 12:23 PM
Subje
That is great, thank you very much David! Yes this seems to work so far :) In
fact I don't have to use //bin/sh . I can just do ahead of time
export MAKEFLAGS='SHELL='
From: David Boyce
It may still be possible to work around these issues via e.g. the environment.
MAKEFLAGS='SHEL
>Yes this seems to work so far :) In fact I don't have to use //bin/sh . I
>can just do ahead of time
export MAKEFLAGS='SHELL='
Until of course, some sub-makefile decided to change MAKEFLAGS internally...
overriding the environment
___
Help-make ma
It's great Paul- thank you! Except, when I do your modification and compile
Make, it then segfaults :)
As my first boss, who was Hawaiian, would say "Auwe!".
Well it looks like I will have to find out why it segfaults...
or maybe try and avoid the "goto" and hope that maybe then the compi
Make 3.81 segfaults if "goto slow" is inserted just before line 2340:
if (shell == 0)
shell = default_shell;
because then, the following line
shell = default_shell;
is not executed and shell is still 0 at (current) line 2710:
bcopy(shell, ap, shell_len);
The next suitable place to
Well, I know I am asking about 3.81 and I am sorry. I don't have a choice,
3.81 is incompatible with 3.82 and forward and we are using 3.81 in that jail.
From: Tim Murphy
To: Mark Galeck
Cc: "psm...@gnu.org" ; "help-make@gnu.org"
Sent: Friday, October 16,
, the argument
(ttyname (fileno (stdout))) has value 0
then this is expected to be a string that later that gets passed to xstrdup and
that segfaults.
From: Tim Murphy
To: Paul D. Smith
Cc: Mark Galeck ; "help-make@gnu.org"
Sent: Tuesday, October 13, 2015 11:04 AM
Subject
Thank you Dan, sorry for being so dumb but... there is no "configure" script
in there, what am I missing??
From: Dan Kegel
Try getting the latest from git; see
http://savannah.gnu.org/git/?group=make
- Dan
___
Help-make mailing list
Help
Yes, I could have read the file README.git after getting the code from git -
perhaps it is obvious that this is the file to read, well it was not obvious to
me.
From: Dan Kegel
To: Mark Galeck
Cc: Tim Murphy ; Paul D. Smith ;
"help-make@gnu.org"
Sent: Friday, October 16,
Dan Kegel wrote:
You may need this commit, which was after 4.1, if I read git history correctly:
commit 292da6f6867b75a5af7ddbb639a1feae022f438f
Author: Paul Smith
Date: Mon Oct 20 01:54:56 2014 -0400
* main.c (main): [SV 43434] Handle NULL returns from ttyname().
Try getting the latest fr
> configure.ac:45: warning: macro `AM_PROG_AR' not found in library
> configure.ac:45: error: possibly undefined macro: AM_PROG_AR
> If this token and others are legitimate, please use m4_pattern_allow.
> See the Autoconf documentation.
> autoreconf: //bin/autoconf failed with exit status: 1
OK I was able to recompile Make 4.1 from git, successfully, without
docs/translations as you have suggested. However, there may be a problem with
the tests. Here is what I did for the record:
install the latest version of gettext GNU package, in non-default location to
overwrite the old insta
>All you have to do is change the list of UNIX shells to be empty:
diff --git a/job.c b/job.c
index 244def0..4b345b6 100644
--- a/job.c
+++ b/job.c
@@ -423,6 +423,7 @@ is_bourne_compatible_shell (const char *path)
??? Forgive me, but I made the change to empty list of unix shell (to the git
Hello,
the GNU Make manual says
(...)executed by invoking a new sub-shell for each line of the recipe, unless
the .ONESHELL special target is in effect (In practice, make may take shortcuts
that do not affect the results.)
Some time ago I was asking here for help in modifying the source of Mak
in the Make manual), but is not a valid shell variable, so the
above construct might not work for GNU Make, period?
I thought the above statement must NOT involve shell at all, precisely
because, it is not a valid shell variable.
Mark
From: Paul Smith
To: Mark Galeck ; "help
Well, I thought I did describe it. I apologize if I was not clear.
the output is
Makefile1:1: VAR-Y
(that is, the value is empty)
From: Philip Guenther
To: Mark Galeck
Cc: "psm...@gnu.org" ; "help-make@gnu.org"
Sent: Friday, April 15, 2016 10:48 PM
Subject: R
> Except by explicit request, `make' exports a variable only if it is
either defined in the environment initially or set on the command line,
and ***if its name consists only of letters, numbers, and underscores.***
Some shells cannot cope with environment variable names consist
Make, which is
different than the parent shell.
Mark
From: Paul Smith
To: Mark Galeck ; "help-make@gnu.org"
Sent: Saturday, April 16, 2016 7:01 AM
Subject: Re: how to (correctly) modify GNU Make so that separate shell is
always spawned for each command
On Sat, 2016-04-1
Hello,
I read and implemented the guidelines in the section of the GNU Make manual
"4.14 Generating Prerequisities Automatically". It works great, for each C
source file, I have a corresponding .d file, which is also a makefile,
generated, and all those makefiles are included in the main makef
Hello,
How can it be - my gnu make is returning 2 but does not show any errors. I am
using -include directive, but I checked and indeed, if make ignores errors
because of -include, then it returns 0, as it should.
So what else could it be? (I can't send you a short example, I can't find the
Hello, I RTFMd gnu make manual carefully, and all I found on how make executes
commands, is that it just passes them to the shell, it does not know anything
about how the shell executes them.
OK, so I am on the Windows XP shell, and I have this makefile:
foobar:
echo %ERRORLE
>I RTFMd gnu make manual carefully
OK, apparently not carefully enough:
5.3.1 Choosing the Shell
says, that /bin/sh will be used by default. If I am on DOS and gnu make is
installed, then likely /bin/sh will be not the DOS shell!
I should say:
SHELL = CMD.EXE
Sorry for the dumb question,
Hello, I have another question, this time I really don't want to look silly,
so I really tried to grep the manual faithfully, and the only place that it
says what happens when you try to kill make, is:
5.6 Interrupting or Killing make
If make gets a fatal signal while a command is executing, it
>we would like some kind of message to the user, indicating that fact, such as
>"nothing to be done for foobar". And that message has to be after the subdir
>messages.
Without that message, the user sees this:
D:\tmp>make foobar
make -C subdir
make[1]: Entering directory `D:/tmp/subdir'
make[
>The sentinel file has to be created: it has to exist. A non-existent
file is always considered out of date. If it exists but then is not
updated, THEN it is not considered out of date.
Yes - I did have it created, and still it did not work, just like I explained.
So, in my last message (be
>All I need is for this example above, to work.
OOPS sorry I made a mistake quoting that example. The bottom makefile is
wrong. Let me try again:
Top makefile is
foobar: | subsystem
echo making foobar
touch foobar
subsystem:
make -C subdi
>That said, if nothing needs to be remade, the effect of recursive make is to
>simply run make several times - one for each sub-make - each of which do
>nothing more than determine that nothing needs to be done.
No, this is my point (I think). The top-level make, will not just "determine
that
>Actually the sentinel is, IMO, an annoying implementation. Managing all
those sentinel files is a PITA. If you're happy to require GNU make,
I'd probably suggest going with order-only prerequisites (or else going
all the way to non-recursive builds) as the most robust solution.
OK, after tryin
t". So my
workaround, was when building the targets, to use a temporary name, and then as
the last operation, rename it to the correct target name. )
Mark
-Original Message-
From: Paul Smith [mailto:psm...@gnu.org]
Sent: Tuesday, November 03, 2009 2:09 PM
To: Mark Galeck (CW)
Cc: Joh
Hello, I see the following recommendation for a "subsystem" recursive make in
the gnu make manual:
subsystem:
$(MAKE) -C subdir
Well, I must not be understanding something, because this does not make (no
pun) sense to me...
The problem I have, is that the "subsystem" target as a
Hello,
Suppose for each .c file such as foobar.c I have generated a makefile foobar.d
file, that lists dependencies of foobar.o on files included in foobar.c.
There are many .d files and each of them has many dependencies, so if I just
blindly do
-include $(OBJS:.o=.d)
then it takes a lot
> See http://make.paulandlesley.org/autodep.html for a good way to handle auto
> dependencies in make. Using the methods described in the article, you will
> only build dependencies when they are needed.
Oh yes, Pete, I am very much aware of this article. In fact, I combined ideas
from that ar
>Hi Paul,
> prog: foosrc/libfoo.a
> ...do something...
>
> foosrc/libfoo.a: FORCE
> $(MAKE) -C foosrc
>
> FORCE: ;
>Sorry if the answer is obvious, but I don't understand why "prog" isn't
>simply always rebuilt, because libfooa is always updated. I
>The time should be relative to number of total .h files (and I guess
number of included files).
Hmm, from what I see, what takes a long time, is all the lines from all the .o
files, to read in all this stuff, is more that a million lines, then I guess
make checks if any of those lines have an
Hello,
The GNU make manual says: " In the VPATH variable, directory names are
separated by colons or blanks. The order in which directories are listed is the
order followed by make in its search."
This does not work for me:
I have a target that depends on several hundred prerequisites, and a
Here is a boiled down example:
Make a subdirectory foobar in your current directory, and put a file foo in
there.
The makefile is
VPATH = \
foobar \
foobar \
foobar \
(... repeated 1000 times)
foo:
Now, if I type
>make foo
On my computer it takes 12 seconds! With VPATH reduced to one foo
Here is another example. I thought that maybe make has some reason to search
all those directories, so I concocted this example. We have 3 subdirectories,
foobar , foobar1, foobar2, in foobar and foobar2 only there is a file foo. The
makefile is:
VPATH = foobar foobar1 foobar1 ... (foobar1 1
Thank you Philip, I was already going to implement my own directory lookup,
because "VPATH is not working". I will follow your suggestions.
>Try doing a make with the -d option and pay close attention to the
long list of patterns that make is applying to find the prerequisites
for all the file
Hello,
When I include submakefiles with
-include foobar
I would like for make to just read the file if there is one, and do nothing
otherwise. Do not consider "foobar" to be a target or anything else. Just
read it if there is one, no other action.
Bart Robinson wrote here that to get of ri
Hello,
I have a target with empty command
.PHONY: foobar
.SILENT: foobar
foobar: ;
When I do
>make foobar
make still says "nothing to be done". How do I arrange the target, so make
does not say anything at all when it encounters it?
Mark
___
Help-m
Hello, I have another, hopefully last one :), question on efficiency, to which
the likely answer maybe "you can't do that".
What I would want, is to set up a rule:
foobar: foobar0 foobar1 foobar2 ... foobar1000
with no commands, all I want is to check, whether file foobar is older than any
o
Never mind, that is not really what is taking that 10 seconds here, sorry for
bothering...
From: help-make-bounces+mgaleck=brocade@gnu.org
[mailto:help-make-bounces+mgaleck=brocade@gnu.org] On Behalf Of Mark Galeck
(CW)
Sent: Friday, November 13, 2009 6:15 PM
To: help-make@gnu.org
>I think your conclusions are a little strong here.
Well yes, I am a perfectionist :) if a tool has a bug, even the smallest one,
that cannot be fixed withing the realm of that tool, then I say it is a flawed
tool.
>Yes, it's true that IF someone adds a new header file that has the same
nam
> is that your performance problems lie elsewhere, other than
make.
Yes you are exactly right, in this case, I screwed up elsewhere. Checking a
thousand files does not take that long.
___
Help-make mailing list
Help-make@gnu.org
http://lists.gnu.o
I want to have a prerequisite file, which may or may not exist. If it exists,
I want make to behave as normal. If it does not exist, I want make to ignore
that prerequisite.
Is it possible?
I tried
foobar: foo
touch foobar
foo:
but here if foo does not exi
Well, thank you for mentioning this David, I will try and work with it. There
is a problem, in that you could have
#include "../foobar.h"
But let me try and see what can be done.
___
Help-make mailing list
Help-make@gnu.org
http://lists.gnu.org/m
>In other words, generate an additional dependency on the DIRECTORIES
in the search path. This works exactly as desired (I think): it
triggers a rebuild whenever a file is added to or deleted from a
directory on the path,
Well, no, I can't do that. The developers will kill me if I do this :) I
Paul wrote:
>Yes, it's true that IF someone adds a new header file that has the same
name as an existing header file, and IF that new header file appears in
a directory that is before the existing header directory on the -I list
for the compiler, and IF the addition of that header file does not
req
>So the "solution" I settled on, is, I checked, that currently in our system no
>file depends on two include files with the same name. Then I will not use
>full-path include dependencies, but only file names:
foobar.o: foobar.h foo.h
Then vpath is split off into a separate makefile, included in
>OK, there is one problem! What if a particularly clever developer, decides to
>not "copy" or make a new file with the same name as the old one "foobar.h".
>But instead, if he does
>mv include1/foo.h include1/foobar.h
Well, I think this is not my problem, but the develo
>
If you mv a file, all files which included the old name must be edited, and
therefore also rebuilt.
---
No, no, it is not foobar.h that is moved. include/foobar.h is included. We
would prepare a new header file, long time ago, with the name include1/foo.h.
And now we
>mv include1/foo
Paul, I appreciate greatly your detailed response and I am very grateful that
you help me in this and other queries. I am right now wearing my C hat and I
don't have time to exactly see if and where I might have gone wrong. Please
forgive me, I will check tomorrow. But maybe I can just show y
Well, I promised not to argue anymore about whether it is bizarre or not so I
will have to uphold my promise unless you release me from it :)
I would still have to tell developers "if you do this bizarre thing, then
rebuild everything", and they would have to check your IFs, let's say I agree
>So, they are going to add a new header file to this directory. Now
EVERY source file that used to include /include1/config.h, and worked fine and
expected that, will start using /include/config.h the next time it's compiled.
>If you have reasonably common situations that meet the criteria, plea
>Mark, I applaud your goal of trying to get make to behave properly in
all cases. However, once you have the -I paths knocked out, does your
approach also work for -L and libraries? What about for programs that
aren't gcc? What happens when you have a Makefile that builds
libfoo.a, but you realize
>I guess here I don't see what the difference is between libraries and
headers. Surely someone could put a generic sounding "libutil.a"
earlier in the library search path. Though probably less likely, you
can rely on the user in one case but not the other?
I always rely on the user, both in case o
>Are you kidding me? You BETTER darn well know. Otherwise, your code
>You guys are talking about an environment that FACILITATES bad coding
practices and ENABLES difficult, time-wasting bugs.
Oh C'mon Paul, I told you many times, I _do_not_ like multiple source files of
the same name, and I a
>In fact, the extra $(wildcard ...) statements have no impact, for good
or ill, on the remove/rename issue. Unless I've confused something
again.
OK, the explanation is not for you Paul - you are the expert :) just a detailed
explanation for other newbies like myself who might stumble upon this.
Hello,
I can use, from inside the makefile, the variable MAKECMDGOALS to find out the
list of goals specified on the make command line. But how do I find out which
goal from that list is the one that is being made currently?
Mark
___
Help-make mailin
>There is no way, unless you want to do something like add an extra
recursion.
Oh I see, thank you Paul, that maybe what I need to do then.
___
Help-make mailing list
Help-make@gnu.org
http://lists.gnu.org/mailman/listinfo/help-make
Hello,
SHELL=cmd.exe
foobar:
@echo.
gives me:
C:\tmp>make foobar
process_begin: CreateProcess(NULL, echo., ...) failed.
make (e=2): The system cannot find the file specified.
make: *** [foobar] Error 2
However, I do use "echo." process all the time on Windows, just outputting to
Oh I see, I thought it must be a "real command", since it worked with the
redirection. Seems like my problem was, I did not know about the "fast path
processing" - I don't see this in the manual, where should I look for this
information?
Thank you very much! Mark
Hmm, are you saying I should report a bug?
>
If it's visible to the user, as it is in this case, then there's a bug
in the fast path processing.
___
Help-make mailing list
Help-make@gnu.org
http://lists.gnu.org/mailman/listinfo/help-make
>make.exe -r
make.exe: Interrupt/Exception caught (code = 0xc0fd, addr = 0x4227d3)
___
Help-make mailing list
Help-make@gnu.org
http://lists.gnu.org/mailman/listinfo/help-make
Well, I found what was causing this. I attempted to call the
$(wildcard ...
function, with about 50,000 arguments (full path names, on average 100
characters, so the total size of all strings that were passed to wildcard, is
about 5MB). Is this a bad idea? :)
>make.exe -r --version
GNU M
>If you're then passing that wildcard list to a program, you could very well be
>exceeding the maximum argument length limit of your shell.
No, I isolated this so there is nothing else, only
foobar: \
$(wildcard \
\
\
(... 5 file paths, average 100 characters each)
\
)
And then
>make
>Probably we use alloca, and this blows up the stack. 50K file names
100 characters each is about 0.5MB.
No... 5MB, I checked the size of the whole thing it's about 5MB.
___
Help-make mailing list
Help-make@gnu.org
http://lists.gnu.org/mailman/list
Hello,
It says in the manual:
$<
The name of the first prerequisite
Is it guaranteed that, if vpath is used, then the directory part of the value
of $< is in fact, one of the vpath paths?
(it could be, that the value of $< were an absolute path for example, while
vpath paths are
Hello, this is a Windows question,
I have 2 subdirectories foobar0 and foobar1, and only in foobar1 there is a
file foobar.s . There is no file foobar.o .
The makefile is:
SHELL=cmd.exe
vpath %.s .\foobar0 .\foobar1
%.o : %.s
echo $<
foobar0\foobar.s
Here's another example, now this one is just terrible, I spent several hours
tracking down this problem:
Again, 2 subdirectories foobar0 and foobar1, and only in foobar1 there is a
file foobar.s . There is no file foobar.o .
The makefile is:
SHELL=cmd.exe
vpath %.s .\foo
Hello, this is about a thread that was here a while ago, where I argued that
some "bizarre" cases should be taken care of by auto-dependency generation.
I want to say, that after chasing this wild goose for a while, I came to
understand that it is not feasible in principle to cover the case o
1 - 100 of 254 matches
Mail list logo
