obstack_printf

2007-04-12 Thread Eric Blake
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 m4 makes heavy uses of obstacks, and I think it is rather wasteful the amount of strings that it puts through asprintf followed by copying to the obstack rather than direct use of obstack_printf. This is because obstack_printf is glibc specific, and n

Re: xnanosleep range with 64 bit time_t

2007-04-12 Thread Eric Blake
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Dmitry V. Levin on 4/12/2007 6:05 PM: > > modules/nanosleep was updated about 2 months ago, but > modules/gettime still depends on itself. Thanks. I fixed it. - -- Don't work too hard, make some time for fun as well! Eric Blake

Re: 'fflush' test failure on Cygwin

2007-04-12 Thread Eric Blake
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Bruno Haible on 4/12/2007 7:30 PM: > > But it's the goal of the 'fflush' module to make this test work, no? Yes > >> But seeing as how the failure symptom is the same as that of MacOSX, > > Indeed, on MacOS X 10.3.9, the lseek call in

Re: seekable stdin test failure on OS X

2007-04-12 Thread Bruno Haible
Eric Blake wrote: > 2007-04-09 Eric Blake <[EMAIL PROTECTED]> > > * modules/fflush: New file. I'm completing the module description like this (for documentation purposes): 2007-04-12 Bruno Haible <[EMAIL PROTECTED]> * modules/fflush (Include): Mention . * modules/strt

Re: 'fflush' test failure on Cygwin

2007-04-12 Thread Bruno Haible
Eric Blake wrote: > > File offset is wrong. > > FAIL: test-fflush.exe > > > > In test-fflush.c line 70, the return value from lseek() is 10 instead of the > > expected 5. > > That is due to a bug in newlib's fflush that I fixed on 2006-12-14. > > I personally don't use a cygwin that old, nor

Re: testing within a git checkout

2007-04-12 Thread Eric Blake
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Bruno Haible on 4/12/2007 4:36 PM: > Eric Blake wrote: >> Would anyone object if I created .cvsignore and .gitignore that ignore the >> pattern testdir*? That way, it would be easier to do in-tree tests with >> './gnulib-tool --with-tests

Re: autobuild.m4

2007-04-12 Thread Paul Eggert
Bruno Haible <[EMAIL PROTECTED]> writes: > if test "$?" != 0 > I thought $? always expands to a non-empty sequence of digits. Are > you expecting a negative integer as value of "$?" ? Even if $? were negative, usages like "test -1 != 0" would work on any host conforming to POSIX, as wel

Re: xnanosleep range with 64 bit time_t

2007-04-12 Thread Dmitry V. Levin
Hi, On Thu, Aug 31, 2006 at 12:13:35AM -0700, Paul Eggert wrote: > I have installed the following patch to gnulib [...] > 2006-08-30 Paul Eggert <[EMAIL PROTECTED]> > > Work around a bug in both the Linux and SunOS 64-bit kernels: > nanosleep mishandles sleeps for longer than 2**31

Re: testing within a git checkout

2007-04-12 Thread Bruno Haible
Eric Blake wrote: > Would anyone object if I created .cvsignore and .gitignore that ignore the > pattern testdir*? That way, it would be easier to do in-tree tests with > './gnulib-tool --with-tests --test ...' without seeing spurious untracked > items when a test fails and leaves behind a directo

Re: autobuild.m4

2007-04-12 Thread Bruno Haible
Hi Simon, > Any objections to this patch? Only minor ones: - About the naming of the naming of the module. It's confusing to see indications that "autobuild" is - A package downloadable from your site that makes it easy to collect the results of automatic builds, - A web site (au

Re: 'fflush' test failure on Cygwin

2007-04-12 Thread Eric Blake
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Bruno Haible on 4/12/2007 2:52 PM: > Hi Eric, > > On Cygwin 2006 (/usr/bin/cygwin1.dll dated 2006-01-20), the test-fflush test > fails: > > File offset is wrong. > FAIL: test-fflush.exe > > In test-fflush.c line 70, the return value

'fflush' test failure on Cygwin

2007-04-12 Thread Bruno Haible
Hi Eric, On Cygwin 2006 (/usr/bin/cygwin1.dll dated 2006-01-20), the test-fflush test fails: File offset is wrong. FAIL: test-fflush.exe In test-fflush.c line 70, the return value from lseek() is 10 instead of the expected 5. Bruno

Re: stdin seekable failure

2007-04-12 Thread Eric Blake
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Eric Blake on 4/12/2007 10:11 AM: >> So I suggest creating a new module 'closein' instead. > > Done like so. > > 2007-04-12 Eric Blake <[EMAIL PROTECTED]> > > Work around glibc's failure to reset seekable stdin on exit. >

Re: stdin seekable failure

2007-04-12 Thread Eric Blake
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Paul Eggert on 4/11/2007 12:41 PM: > Eric Blake <[EMAIL PROTECTED]> writes: > >> Maybe the closeout module needs to be updated to worry about stdin as well >> as stdout/stderr? > > Coreutils doesn't need that, as it doesn't have the prob

Re: autobuild.m4

2007-04-12 Thread Simon Josefsson
Eric Blake <[EMAIL PROTECTED]> writes: > According to Simon Josefsson on 4/12/2007 6:51 AM: >> >> I just realized a namespace problem, when I wanted to apply the >> following obvious patch: >> >> +configure.ac-early: >> +AC_REQUIRE([AB_INIT]) >> + > > Currently, the instructions at > http://jose

Re: autobuild.m4

2007-04-12 Thread Eric Blake
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Simon Josefsson on 4/12/2007 6:51 AM: > > I just realized a namespace problem, when I wanted to apply the > following obvious patch: > > +configure.ac-early: > +AC_REQUIRE([AB_INIT]) > + Currently, the instructions at http://josefsson.o

Fix --create-megatestdir

2007-04-12 Thread Simon Josefsson
Creating a script called 'autobuild' doesn't work well when there is a directory called 'autobuild'... ? MODULES.html Index: ChangeLog === RCS file: /sources/gnulib/gnulib/ChangeLog,v retrieving revision 1.1557 diff -u -p -r1.1557 Cha

Fix to make sysexits work

2007-04-12 Thread Simon Josefsson
Building 'sysexits' seems to have stopped working, this fixed it. Index: ChangeLog === RCS file: /sources/gnulib/gnulib/ChangeLog,v retrieving revision 1.1556 diff -u -p -r1.1556 ChangeLog --- ChangeLog 12 Apr 2007 12:02:23 -

Re: autobuild.m4

2007-04-12 Thread Simon Josefsson
Simon Josefsson <[EMAIL PROTECTED]> writes: >> Also, is the master of this file still autobuild (in which case Karl >> would want to set up an auto-syncer), or do you intend for autobuild >> to now pull the latest rev from gnulib? > > It seems simpler to have the master be gnulib. I'm pulling > a

Re: autobuild.m4

2007-04-12 Thread Simon Josefsson
Eric Blake <[EMAIL PROTECTED]> writes: > According to Simon Josefsson on 4/12/2007 5:20 AM: >> Eric Blake <[EMAIL PROTECTED]> writes: >> >>> According to Simon Josefsson on 4/12/2007 2:31 AM: Several projects uses autobuild.m4 (at least gnutls, gsasl, inetutils, libidn) and also use gnu

Re: seekable stdin test failure on OS X

2007-04-12 Thread Eric Blake
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Bruno Haible on 4/10/2007 4:29 PM: > Hi Eric, > > The 'e1' variable is not needed, I think. No standard specifies that errno > must be preserved across function calls: > - C99 does not mention 'errno' in the description of fflush, > -

Fix to make --avoid=dummy work

2007-04-12 Thread Simon Josefsson
I have installed this -- autobuild is a shell-script-only project but it uses gnulib for the 'autobuild.m4' file. /Simon ? MODULES.html Index: ChangeLog === RCS file: /sources/gnulib/gnulib/ChangeLog,v retrieving revision 1.1553 diff

Re: autobuild.m4

2007-04-12 Thread Eric Blake
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Simon Josefsson on 4/12/2007 5:20 AM: > Eric Blake <[EMAIL PROTECTED]> writes: > >> According to Simon Josefsson on 4/12/2007 2:31 AM: >>> Several projects uses autobuild.m4 (at least gnutls, gsasl, inetutils, >>> libidn) and also use gnu

testing within a git checkout

2007-04-12 Thread Eric Blake
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Would anyone object if I created .cvsignore and .gitignore that ignore the pattern testdir*? That way, it would be easier to do in-tree tests with './gnulib-tool --with-tests --test ...' without seeing spurious untracked items when a test fails and le

Re: autobuild.m4

2007-04-12 Thread Simon Josefsson
Eric Blake <[EMAIL PROTECTED]> writes: > According to Simon Josefsson on 4/12/2007 2:31 AM: >> Several projects uses autobuild.m4 (at least gnutls, gsasl, inetutils, >> libidn) and also use gnulib. It seems easier to distribute >> autobuild.m4 via gnulib. Any objections to this patch? > > Sounds

Re: autobuild.m4

2007-04-12 Thread Eric Blake
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Simon Josefsson on 4/12/2007 2:31 AM: > Several projects uses autobuild.m4 (at least gnutls, gsasl, inetutils, > libidn) and also use gnulib. It seems easier to distribute > autobuild.m4 via gnulib. Any objections to this patch? Sounds

Re: seekable stdin test failure on OS X

2007-04-12 Thread Gary V. Vaughan
Hi Eric, [for bug-gnulib, this is on Mac OS 10.4.9 intel] On 11 Apr 2007, at 13:39, Eric Blake wrote: gnulib-tool --with-tests --test fflush make check-TESTS File offset is wrong. FAIL: test-fflush PASS: test-stdio PASS: test-unistd === 1 of 3 tests failed ==

autobuild.m4

2007-04-12 Thread Simon Josefsson
Several projects uses autobuild.m4 (at least gnutls, gsasl, inetutils, libidn) and also use gnulib. It seems easier to distribute autobuild.m4 via gnulib. Any objections to this patch? /Simon Index: ChangeLog === RCS file: /sources