Rusty Russell [Tue, 26 Jun 2012 06:46:18 +0000 (16:16 +0930)]
bytestring: new module.
Alessandro Guido [Wed, 20 Jun 2012 08:46:19 +0000 (10:46 +0200)]
Avoid compiler errors if u32 is defined as a type
Fixes this error:
/hash.h:304:13: error: two or more data types in
declaration specifiers
Rusty Russell [Sat, 9 Jun 2012 06:06:45 +0000 (15:36 +0930)]
antithread, failtest: use ccan/err instead of err.h.
Rusty Russell [Sat, 9 Jun 2012 06:04:56 +0000 (15:34 +0930)]
opt: avoid using err.h.
Seems it's not available on Solaris.
Rusty Russell [Fri, 8 Jun 2012 11:19:32 +0000 (20:49 +0930)]
err: actually implement replacement versions!
This is where you see the genius of the originals: you can't implement them
without horrible hacks to get the program name :(
Rusty Russell [Fri, 8 Jun 2012 10:58:01 +0000 (20:28 +0930)]
Merge branch 'master' of ozlabs.org:ccan
Rusty Russell [Fri, 8 Jun 2012 07:51:04 +0000 (17:21 +0930)]
configurator: add HAVE_ERR_H test.
Rusty Russell [Fri, 8 Jun 2012 07:37:03 +0000 (17:07 +0930)]
ccan/err: new err.h-replacing module.
Seems like Solaris doesn't have err.h, as discovered by Samba.
Michael Ellerman [Fri, 25 May 2012 06:38:08 +0000 (16:38 +1000)]
tools/create-ccan-tree: Add basic waf support
This adds a "waf" buildtype to create-ccan-tree. When passed it
generates a basic wscript for building everything into a libccan.a
It doesn't support running the generator, so you need to provide
config.h
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeremy Kerr <jeremy.kerr@canonical.com>
Jeremy Kerr [Thu, 24 May 2012 05:48:19 +0000 (13:48 +0800)]
tools/create-ccan-tree: Add --build-type=automake
Add a new build type to create an appropriate Makefile.am for the
newly-created ccan tree.
Signed-off-by: Jeremy Kerr <jeremy.kerr@canonical.com>
Jeremy Kerr [Thu, 24 May 2012 01:48:00 +0000 (09:48 +0800)]
tools/create-ccan-tree: Add --build-type=TYPE
Replace the --exclude-configurator argument with a
single --build-type=TYPE argument, allowing future expansion of build
system types (and allowing source-tree-only creation). We currently
support:
* 'make' (just the Makefile, equivalent to --exclude-configurator)
* 'make+config' (current default)
Signed-off-by: Jeremy Kerr <jeremy.kerr@canonical.com>
Jeremy Kerr [Thu, 24 May 2012 03:49:10 +0000 (11:49 +0800)]
tools/create-ccan-tree: replace --exclude-tests with --copy-all option
Rather than explicitly excluding the tests, just copy the source files
by default. Add an option (--copy-all) to include all of the non-source
stuff in each module.
Signed-off-by: Jeremy Kerr <jeremy.kerr@canonical.com>
Jeremy Kerr [Tue, 22 May 2012 09:19:24 +0000 (17:19 +0800)]
tools: ccan_depends requires a config.h
Commit
a77cc282 causes tools/create-ccan-tree to fail, as it can't build
tools/ccan_depends (via `make tools/ccan_depends`) without a
configuration header present.
This change adds a dependency on config.h, so that the configurator
kicks-in prior to building ccan_depends.o.
Other tools don't need this dependency, as tools/ccan_depends is built
during the Makefile 'include' process.
Signed-off-by: Jeremy Kerr <jeremy.kerr@canonical.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Rusty Russell [Thu, 17 May 2012 03:44:18 +0000 (13:14 +0930)]
config.h: don't include in git repository.
We create and upload an example for the web anyway, if people really need it.
Jeremy Kerr [Wed, 16 May 2012 03:02:56 +0000 (11:02 +0800)]
tools: Add create-ccan-tree utility
Add a script to create a ccan source tree for inclusion into projects'
distributed code.
create-ccan-tree takes a list of ccan modules and copies them, plus any
dependent modules, into a target directory. Enough build infrastructure
is added to build libccan.a from the ccan modules.
Typical usage for my project that uses talloc and read-write-all:
lib/ccan.git/tools/create-ccan-tree --exclude-tests \
lib/ccan/ talloc read-write-all
- this uses a git submodule in $PWDlib/ccan.git to create a source tree
in $PWD/lib/ccan, containing code for talloc and read-write-all.
Signed-off-by: Jeremy Kerr <jeremy.kerr@canonical.com>
Rusty Russell [Thu, 29 Mar 2012 04:28:33 +0000 (14:58 +1030)]
asort: fix test when typesafe_cb not supported.
Solaris CC seems to support typeof, but not __builtin_types_compatible_p.
Rusty Russell [Thu, 29 Mar 2012 04:28:33 +0000 (14:58 +1030)]
cast: make sure suncc sees a constant.
cast_const() et. al. are supposed to be a constant expression, so you can do things like:
static char *p = cast_const(char *, (const char *)"hello");
Unfortunately, a cast to intptr_t and arithmetic makes suncc reject it as
a constant expression. We need the cast, because (1) the expression could be
a void *, so we can't just add to it, and (2) gcc complains with -Wcast-qual
without it.
So instead of adding BUILD_BUG_OR_ZERO, we use a ? :, which keeps everyone happy.
Rusty Russell [Thu, 29 Mar 2012 04:28:23 +0000 (14:58 +1030)]
ccan_tokenizer: don't assume __VA_ARGS__ can be empty.
That's a gcc extension.
Rusty Russell [Tue, 27 Mar 2012 05:10:45 +0000 (15:40 +1030)]
cast: test/compile_fail-cast_static.c should fail without COMPOUND_LITERALS.
It still gave a warning on gcc, because casting a char to a char* gives a warning. Not so on sun CC.
Rusty Russell [Tue, 27 Mar 2012 05:01:27 +0000 (15:31 +1030)]
ccanlint: module_links test must link with module.
Gcc doesn't include unused inline functions, even without -O, but Solaris CC
does, so fails to link block_pool where one inline function refers to a
non-inline function.
Rusty Russell [Tue, 27 Mar 2012 04:37:50 +0000 (15:07 +1030)]
asprintf: #include <stdarg.h> in test/run.c
We get away with it on Linux, not on Solaris.
Rusty Russell [Tue, 27 Mar 2012 04:36:46 +0000 (15:06 +1030)]
ccanlint: tests_compile_coverage needs GCC.
We only support gcc/gcov so far.
Rusty Russell [Tue, 27 Mar 2012 04:22:13 +0000 (14:52 +1030)]
ccanlint: add ilog dependency.
This only matters on non-gcc, where ilog functions are not just mapped to builtins.
Rusty Russell [Tue, 27 Mar 2012 04:08:13 +0000 (14:38 +1030)]
tap: assume variadic macros.
The Sun C compiler has them, but fails this overzealous check.
Rusty Russell [Tue, 27 Mar 2012 04:00:29 +0000 (14:30 +1030)]
opt: fix Solaris compile.
Need sys/termios.h for struct winsize.
Rusty Russell [Tue, 27 Mar 2012 03:54:00 +0000 (14:24 +1030)]
iscsi: use <sys/filio.h> on Solaris.
Needed for FIONREAD.
Rusty Russell [Tue, 27 Mar 2012 03:45:19 +0000 (14:15 +1030)]
alloc: fix return of void.
Sun CC complains.
Rusty Russell [Tue, 27 Mar 2012 03:45:18 +0000 (14:15 +1030)]
tools: don't use __attribute__
We have compiler.h for that.
Rusty Russell [Tue, 27 Mar 2012 03:45:18 +0000 (14:15 +1030)]
tools: fix bogus unreachable return.
Rusty Russell [Tue, 27 Mar 2012 03:45:18 +0000 (14:15 +1030)]
tools/configurator: fix test for /proc/self/maps.
Header file was wrong, suncc gives warning about missing open prototype, causing configurator to fail.
Rusty Russell [Tue, 27 Mar 2012 03:45:16 +0000 (14:15 +1030)]
Makefile: generate config.h in two stages
This means we don't create an empty file if configurator fails.
Rusty Russell [Wed, 21 Mar 2012 00:48:18 +0000 (11:18 +1030)]
failtest: compile fix for OpenBSD
OpenBSD doesn't idempotent-wrap sys/mman.h, so when we #define mmap to
an alternative, it fails to compile when sys/mman.h is included again.
Workaround is not to #define mmap to add arguments on Open BSD.
Rusty Russell [Mon, 19 Mar 2012 05:51:00 +0000 (16:21 +1030)]
Merge branch 'ccantool'
Rusty Russell [Mon, 19 Mar 2012 05:36:27 +0000 (16:06 +1030)]
tools: move config.h reading to its own file
Move ccan_dir determination out to its own function, rather than implying it
by the first time we get the manifest of a module.
Rusty Russell [Mon, 19 Mar 2012 05:15:45 +0000 (15:45 +1030)]
tools: explicit find_ccan_dir()
Move ccan_dir determination out to its own function, rather than implying it
by the first time we get the manifest of a module.
Rusty Russell [Mon, 19 Mar 2012 05:14:59 +0000 (15:44 +1030)]
ccanlint: handle -x correctly with multiple modules.
After the first module we were resetting even the excluded tests, which caused
a segfault as they weren't initialized.
Rusty Russell [Mon, 19 Mar 2012 05:14:13 +0000 (15:44 +1030)]
tools: determing ccan directory differently
Don't assume it's two levels above the module dir, instead look for
last "/ccan/" in path. This is an incremental step towards nested
module directories.
Rusty Russell [Mon, 19 Mar 2012 05:13:13 +0000 (15:43 +1030)]
tools: split out manifest from ccanlint
Other tools will soon want to use this, so make it independent.
(eg. ccantool).
Rusty Russell [Fri, 9 Mar 2012 02:55:03 +0000 (13:25 +1030)]
tdb2: remove: it's now in SAMBA where it belongs.
Rusty Russell [Fri, 9 Mar 2012 02:50:35 +0000 (13:20 +1030)]
htable: fix bug where first entry has hash of 0 or 1.
Thanks to Zoltán Lajos Kis for the bug report and test case!
Rusty Russell [Thu, 8 Mar 2012 03:44:22 +0000 (14:14 +1030)]
failtest: don't assume FD_SETSIZE is maximum runtime fd.
This breaks when rlimit is less. Unfortunately, valgrind (32 bit x86,
3.7.0.SVN, Ubuntu) fails to set the file limit properly on the test:
reducing it to the obvious getrlimit/setrlimit/getrlimit works fine,
so leaving diagnostics for another day.
Daniel Burke [Sun, 26 Feb 2012 17:03:59 +0000 (03:33 +1030)]
ttxml: made alloc failure consistent - return NULL
Daniel Burke [Sun, 26 Feb 2012 17:03:05 +0000 (03:33 +1030)]
ttxml: removed cruft from tests
Daniel Burke [Sun, 26 Feb 2012 17:02:39 +0000 (03:32 +1030)]
ttxml: fixed license test
Rusty Russell [Tue, 14 Feb 2012 09:39:18 +0000 (20:09 +1030)]
tdb: delete from CCAN.
This was an early experiment in putting code into CCAN, but TDB is a public
library distributed as part of the Samba sources; there is no reason to
publish it here (especially now the unit tests are also in Samba).
Rusty Russell [Tue, 14 Feb 2012 09:34:00 +0000 (20:04 +1030)]
configurator: define HAVE_CCAN.
An interesting case came up with the tdb2 code in SAMBA recently. It's a
public library, and it doesn't want to be dependent on CCAN, but the header
uses cast and typesafe_cb for extra type safety.
A good solution to this is to put dummy versions under !HAVE_CCAN. Of course,
any CCAN config.h should define HAVE_CCAN.
Rusty Russell [Sat, 28 Jan 2012 04:14:43 +0000 (14:44 +1030)]
tdb2: copy tdb1's changed expansion logic.
TDB2 uses the same expansion logic as TDB1, which got factored out
recently. So update TDB2 to match.
Rusty Russell [Sat, 28 Jan 2012 04:14:43 +0000 (14:44 +1030)]
tdb2: tdb1: use same expansion factor logic when expanding for new recovery area.
If we're expanding because the current recovery area is too small, we
expand only the amount we need. This can quickly lead to exponential
growth when we have a slowly-expanding record (hence a
slowly-expanding transaction size).
(Syncs with tdb1 SAMBA patch
3a2a755e3380a8f81374009d463cd06161352507)
Volker Lendecke [Sat, 28 Jan 2012 04:14:36 +0000 (14:44 +1030)]
tdb2: Avoid a malloc/memcpy in _tdb1_store
(Syncs tdb2 with SAMBA tdb1's
664add17757836c5ee98618aef11371f412b6e44)
Rusty Russell [Wed, 21 Dec 2011 05:54:41 +0000 (16:24 +1030)]
tdb2: careful on wrap.
It's much harder to wrap a 64-bit tdb2 than a 32-bit tdb1, but we should still
take care against bugs.
Also, we should *not* cast the length to a size_t when comparing it to
the stat result, in case size_t is 32 bit.
Rusty Russell [Wed, 21 Dec 2011 05:44:51 +0000 (16:14 +1030)]
tdb2: be more careful on 4G files (tdb1).
I came across a tdb which had wrapped to 4G + 4K, and the contents had been
destroyed by processes which thought it only 4k long. Fix this by checking
on open, and making tdb_oob() check for wrap itself.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
(Ported from tdb1 SAMBA commit
b64494535dc62f4073fc6302847593ed6e6ec38b)
Rusty Russell [Tue, 20 Dec 2011 07:10:57 +0000 (17:40 +1030)]
tdb: don't free old recovery area when expanding if already at EOF.
We allocate a new recovery area by expanding the file. But if the
recovery area is already at the end of file (as shown in at least one
client case), we can simply expand the record, rather than freeing it
and creating a new one.
Rusty Russell [Tue, 20 Dec 2011 07:09:28 +0000 (17:39 +1030)]
tdb: use same expansion factor logic when expanding for new recovery area.
If we're expanding because the current recovery area is too small, we
expand only the amount we need. This can quickly lead to exponential
growth when we have a slowly-expanding record (hence a
slowly-expanding transaction size).
Rusty Russell [Tue, 20 Dec 2011 07:07:45 +0000 (17:37 +1030)]
tdb: test exponential recovery expansion problem.
Rusty Russell [Mon, 19 Dec 2011 01:21:45 +0000 (11:51 +1030)]
opt: much prettier usage (using terminal size)
Rusty Russell [Fri, 16 Dec 2011 00:09:47 +0000 (10:39 +1030)]
opt: handle -Wcast-qual warnings.
Rusty Russell [Thu, 15 Dec 2011 03:45:47 +0000 (14:15 +1030)]
opt: don't leak on exit (valgrind complains).
Recent real usage case showed this leak when we call opt_usage_and_exit:
we don't bother freeing before exit. With valgrind, it matters.
Andrey Smirnov [Mon, 12 Dec 2011 03:22:44 +0000 (13:52 +1030)]
list.h: opaque list iteration functionality
See https://github.com/rustyrussell/ccan/pull/1#issuecomment-
2856561
for details.
(Reworked with minor cleanups -- Rusty)
Rusty Russell [Mon, 12 Dec 2011 03:20:05 +0000 (13:50 +1030)]
container_of: add container_off_var macro
Based on patch by Andrey Smirnov <andrew.smirnov@gmail.com>:
There is are certain use-cases when it is necessary to know the offset
of the member in a structure's memory layout. One such use-case can be
seen in `ccan/list/list.h' in macros `list_for_each' and
`list_for_each_safe'. This commit implements said functionality with
`container_of_var_off' macro.
Renamed container_of_var_off -> container_off_var now we have container_off.
Rusty Russell [Mon, 12 Dec 2011 03:20:05 +0000 (13:50 +1030)]
check_type: fix incorrect documentation.
Rusty Russell [Mon, 12 Dec 2011 03:20:05 +0000 (13:50 +1030)]
container_of: don't put member_ptr in container_off.
It's convenient to check that the member is the given type, but we can leave
that to the callers.
Andrey Smirnov [Mon, 12 Dec 2011 03:19:57 +0000 (13:49 +1030)]
Makefile, configurator: Add additional debug flags to CFLAGS
Just adding `-g' to list of CFLAGS doesn't make gcc to generate debug
information required for macro expansion during debugging. Replacing
it with `-g3 -ggdb' rectifies this.
(configurator.c commit folded by Rusty)
Rusty Russell [Wed, 7 Dec 2011 02:57:27 +0000 (13:27 +1030)]
opt: suggest they look at ccan/autodata.
Nice for distributed option declaration.
Rusty Russell [Wed, 7 Dec 2011 02:57:14 +0000 (13:27 +1030)]
autodata: add example, clean up description a little.
Also add #include to AUTODATA_TYPE example, so ccanlint doesn't try to
put it inside a function (which works, for this, but gives a warning
about unused variables).
Rusty Russell [Wed, 7 Dec 2011 02:55:41 +0000 (13:25 +1030)]
tools/doc_extract: trim trailing blank lines on sections.
Makes things a bit neater.
Rusty Russell [Wed, 7 Dec 2011 02:55:41 +0000 (13:25 +1030)]
web: publish See Also: for modules which supply it.
Rusty Russell [Wed, 7 Dec 2011 02:55:41 +0000 (13:25 +1030)]
tools/doc_extract: simplify handling fields with spaces in them.
eg. "See also" can match "see-also".
Rusty Russell [Mon, 5 Dec 2011 12:48:50 +0000 (23:18 +1030)]
ccanlint: use ccan/autodata
Gets rid of generated file.
Rusty Russell [Mon, 5 Dec 2011 12:48:21 +0000 (23:18 +1030)]
autodata: stash pointers in a binary.
This is a more portable variant of the ELF section trick.
Rusty Russell [Mon, 5 Dec 2011 12:17:46 +0000 (22:47 +1030)]
configurator: HAVE_SECTION_START_STOP
Rusty Russell [Mon, 5 Dec 2011 12:16:46 +0000 (22:46 +1030)]
ptr_valid: test whether a ptr is valid.
Very slow, but sometimes you need to know without crashing.
Rusty Russell [Mon, 5 Dec 2011 03:46:33 +0000 (14:16 +1030)]
configurator: HAVE_PROC_SELF_MAPS
Rusty Russell [Sun, 4 Dec 2011 22:50:34 +0000 (09:20 +1030)]
objset: new module.
Rusty Russell [Sun, 4 Dec 2011 22:49:34 +0000 (09:19 +1030)]
ccanlint.1: fix description of per-test options for tests_pass_valgrind.
Rusty Russell [Fri, 2 Dec 2011 03:15:08 +0000 (13:45 +1030)]
doc: man page for ccanlint.
Rusty Russell [Fri, 2 Dec 2011 03:37:24 +0000 (14:07 +1030)]
ccanlint: remove argument to -k/--keep
It's much easier to simply say "keep all", and it simplifies the code
quite a bit.
Rusty Russell [Fri, 2 Dec 2011 03:37:23 +0000 (14:07 +1030)]
ccanlint: use -t for --target not --timeout
I use --target far more than --timeout.
Rusty Russell [Fri, 2 Dec 2011 03:37:22 +0000 (14:07 +1030)]
ccanlint: use traversal to trim tests we want to skip.
Rather than using a separate index of tetst to skip, make sure tests
we don't want to run are removed from the graph of tests to start
with.
This means that the "skip" field is only set when we fail a test (and
thus must be reset when we test the next module).
Rusty Russell [Fri, 2 Dec 2011 03:37:21 +0000 (14:07 +1030)]
ccanlint: use node traverse for --target.
Rather than walking the list of tests manually, use
dgraph_traverse_to() to run the tests.
Rusty Russell [Fri, 2 Dec 2011 03:37:21 +0000 (14:07 +1030)]
ccanlint: use dgraph module.
Instead of a linked list of tests, we use dgraph and strmap.
Rusty Russell [Fri, 2 Dec 2011 03:37:20 +0000 (14:07 +1030)]
ccanlint: remove wart from info options.
Make info_exists() call add_info_options() directly, rather than detecting
that we've found _info inside the loop of tests.
It means targeting is now a global, which is a bit sucky.
Rusty Russell [Fri, 2 Dec 2011 03:36:58 +0000 (14:06 +1030)]
dgraph: new module for directed graphs.
Rusty Russell [Fri, 2 Dec 2011 03:10:27 +0000 (13:40 +1030)]
ccanlint: parse --verbose before anything else.
This way, we can debug really early failures (eg. during option
parsing).
Rusty Russell [Thu, 1 Dec 2011 06:15:51 +0000 (16:45 +1030)]
ccanlint: get rid of separate class of compulsory tests.
It's reduced to a flag which means we stop processing the module. We have
to enhance our dependencies a bit, but they've always been quite good.
This simplifies things quite a bit.
Rusty Russell [Thu, 1 Dec 2011 06:14:51 +0000 (16:44 +1030)]
opt: add OPT_EARLY and opt_early_parse.
Parsing options like --verbose and --debug can be a pain. You need to
have everything set up before invoking parse_args(), but that may be a
significant amount of work, for which you may want verbose or
debugging enabled.
Thus the concept of "early" args: you can nominate arguments to be
parse before anything else, using opt_early_parse().
Rusty Russell [Thu, 1 Dec 2011 06:13:51 +0000 (16:43 +1030)]
strset, strmap: invert iterator function meaning.
Make a false return abort the iteration, not true.
The old way makes sense for search functions (true == I found it), but
other kinds of iteration are more common (brute force search is
probably dumb).
Rusty Russell [Thu, 1 Dec 2011 06:12:51 +0000 (16:42 +1030)]
setset: fix API to match strmap and common sense.
strset_set -> strset_add
strset_clear -> strset_del
strset_test -> strset_get
strset_destroy -> strset_clear
Rusty Russell [Thu, 1 Dec 2011 06:11:51 +0000 (16:41 +1030)]
strmap: allow const arguments to strset_iterate().
Rusty Russell [Thu, 1 Dec 2011 06:10:51 +0000 (16:40 +1030)]
strset: allow const arguments to strset_iterate().
Rusty Russell [Thu, 1 Dec 2011 06:09:51 +0000 (16:39 +1030)]
opt: fix up outdated comments in documentation.
Rusty Russell [Thu, 1 Dec 2011 06:09:50 +0000 (16:39 +1030)]
opt: get rid of last remnant of getopt.
Testing code still using it.
Rusty Russell [Thu, 1 Dec 2011 06:09:47 +0000 (16:39 +1030)]
tlist: remove type arg from tlist_top(), tlist_tail()
With the type canary, it's unnecessary. Though the implementation is
a bit more awkward since they longer map directly through to list_top/tail.
Rusty Russell [Wed, 30 Nov 2011 03:02:18 +0000 (13:32 +1030)]
list: don't multiple-evaluate arguments to tlist_top and tlist_tail
Rusty Russell [Wed, 30 Nov 2011 03:02:06 +0000 (13:32 +1030)]
container_of: add container_off() helper.
Just like container_of() (with typechecking, etc), but return the offset
to the enclosing structure.
Rusty Russell [Wed, 30 Nov 2011 02:41:00 +0000 (13:11 +1030)]
ttxml: exclude license_comment check so it doesn't fail ccanlint.
Rusty Russell [Wed, 30 Nov 2011 02:39:07 +0000 (13:09 +1030)]
tdb2: display capability information in tdb_summary()
This means we know they're there in future, and what restrictions they
carry.
Rusty Russell [Wed, 30 Nov 2011 01:39:18 +0000 (12:09 +1030)]
tdb2: add a capability list from the header.
This allows even more extensibility in future: in particular, the top
bits of each capability tell us what to do if we don't understand it:
fail the open, fail to open for write, or don't try to check the
format.
tdb_check needs to understand the capability list so it can know to
skip over it: each element in the list is prefixed with the type tag
and the length.
Rusty Russell [Wed, 30 Nov 2011 00:42:22 +0000 (11:12 +1030)]
tdb2: provide tdb_layout_write() rather than implying it by new_tdb_layout arg.
Neater API.
Rusty Russell [Wed, 30 Nov 2011 00:42:22 +0000 (11:12 +1030)]
tdb2: add an internal TDB_CANT_CHECK flag.
This will be used shortly to indicate that a TDB2 file indicates it
cannot be checked.
Rusty Russell [Wed, 30 Nov 2011 00:42:03 +0000 (11:12 +1030)]
failtest: do trace via a FILE, not an fd.
The current mix of writing to an fd doesn't mix as well with
stderr/stdout. And writing to a FILE * is more portable.