more globally, since the 2006-06-30 patch didn't suffice. Problem
reported by Keith Marshall. Also, don't bother with builddir2,
since it shouldn't be needed any more.
"HP" and "HP-UX". Remove mention of OSF; the old version wasn't
quite right (it talked about "OSF/Tru64", even though the
operating systems were called OSF/1, Digital UNIX, and Tru64 UNIX,
and it even mentioned "OSF 4"!) and at this point there's little
reason to talk about OSF any more, since it died in 1994.
(Specific Compiler Characteristics): Simplify example of
negative-size array.
(File Descriptors): Reorder to make the text flow better.
Remove joke about "appreciate the various levels"; I didn't get it.
Add remark about HP-UX sh -x bug with stderr noted by Bob Proulx in
<http://lists.gnu.org/archive/html/bug-coreutils/2006-06/msg00225.html>.
(File Descriptors, Limitations of Usual Tools):
Tone down the advice against renaming or removing open files.
(Limitations of Usual Tools): Add a new section, on 'rm'.
datarootdir-related errors only if AC_DATAROOTDIR_CHECKED is
not defined.
* doc/autoconf.texi (Changed Directory Variables): New node,
to document the whole `datarootdir' business a bit better.
* NEWS: Update.
* tests/torture.at (datarootdir workaround): Extend test.
Prompted by report by Alexandre Julliard.
Problem reported by Eric Blake.
* doc/autoconf.texi: Likewise.
Use @acronym around BSD, GCC, and GNU when appropriate.
(Why GNU M4): Renamed from "Why GNU m4".
(Redefined M4 Macros): Mention that Posix
m4wrap takes only 1 argument, but GNU M4 1.4.x takes more.
(Buffer Overruns): Mention size_t and ptrdiff_t as alternatives
to int.
calls, so that we do not care whether they are LIFO or FIFO;
in the m4_wrap, do not check which diversion is the topmost
one, just check that the stack is balanced at the end.
* lib/m4sugar/m4sh.m4 (AS_INIT): We are going to change the
base diversion forever--pop the previous diversion before
opening the new one; consequently, remove the m4_wrap call.
* lib/autotest/general.m4 (AT_INIT): Likewise.
* tests/m4sugar.at: Do not use
m4_wrap([m4_diversion_pop([..])]), for educational purposes.
* doc/autoconf.texi (): Likewise.
Rework example of m4wrap token-pasting trouble so that it doesn't
care whether it's LIFO or FIFO.
Fix some "contrary to"s that are awkward in English.
to `yes' instead of `int$1_t' if the type is found, for more
consistent configure output (where $1 is the number of bits).
(_AC_TYPE_UINT): Likewise for `uint$1_t'.
Suggested by Bruno Haible.
in order to avoid picking up an older installed frozen m4sh.m4f.
Besides an outdated shell startup, this could have been created
by an earlier M4 version with incompatible frozen file format.
(Installation Directory Variables, Build Directories):
(Automatic Remaking, Subdirectories, Fortran Compiler):
(Making testsuite Scripts, Defining Directories):
Quote variable usages better.
(Installation Directory Variables): Fix table item font.
Reword slightly to clarify. Generalize advice about
not using special characters to include all file-related
vars, not just VPATH.
(Special Chars in Variables): Warn about special characters in
$(srcdir) too.
(Assignments): Clarify default-value example as suggested by
Ralf Wildenhues in
<http://lists.gnu.org/archive/html/autoconf-patches/2006-06/msg00072.html>.
(Special Shell Variables): Note leading ./ or ../, as suggested
by Stepan Kasal.
(Limitations of Builtins): Under cd, warn about CDPATH.
(The Make Macro MAKEFLAGS): Untabify. Problem reported by
Ralf Wildenhues.
many suggested by Ralf Wildenhues.
Reword to avoid "@code{...}'s" and the like, since it's ugly
with Emacs info mode. discontents -> woes.
Put a few "will"s back. time stamp -> timestamp.
side-effect -> side effect.
follows. Use "makefile" when talking about makefiles
generally (which might be named "makefile" or "Makefile" or even
"foo.mk"), "Makefile" when talking about a specific makefile
called "Makefile". This unclutters the text from weird quotes
(e.g., "`Makefile's" in info mode). Similarly, use "@var{foo}
values" rather than "@var{foo}s" and similar constructs containing
"}s". Use "Make rules" rather than "Makefile rules". Minor
English-language improvements. Change the prefix "sub-" to "sub"
and "re-" to "re".
Put blank lines around examples more consistently.
Avoid "rather" and "very" as intensifiers.
Avoid "will" as an auxiliary.
(Limitations of Make): Split this node into....
(Portable Make, $< in Ordinary Make Rules, Failure in Make Rules):
(Leading _ in Macro Names, Backslash-Newline-Newline):
(Backslash-Newline Comments, Long Lines in Makefiles):
(Macros and Submakes, The Make Macro SHELL, Comments in Make Rules):
(obj/ and Make, make -k Status, VPATH and Make):
(VPATH and Double-colon, $< in Explicit Rules):
(Automatic Rule Rewriting, OSF/Tru64 Directory Magic):
(Make Target Lookup, Single Suffix Rules, Timestamps and Make):
New nodes, resulting from splitup of Limitations of Make.
All cross-references changed. Raise the top node from
a section to a chapter, and all subnodes accordingly.
Redo the introductory wording to match the new organization.
(Installation Directory Variables): Use an example that is
closer to what Autoconf actually does. Mention that VPATH's
value should not contain metacharacters or white space.
(Fortran Compiler): Fix a VPATH bug in an example.
(Leading _ in Macro Names): Mention that this problem is no longer
of practical concern.
(VPATH and Make): Reword the advice to make it clearer
that Autoconf and Automake support VPATH in non-GNU make, but
many packages have bugs in this area.
($< in Explicit Rules): Refer to Build Directories rather
than using a (non-VPATH-safe) example.
(Automatic Rule Rewriting): Mention the sort of disaster that
can ensue with Solaris-style rule rewriting with VPATH.
