Re: getloadavg test

2011-02-17 Thread Paul Eggert
On 02/16/2011 03:28 AM, Bruno Haible wrote: How about this? It's basically what a human would do when executing the manual test. Sure, something like that sounds fine. Though I'd make the upper bound higher than 100. Maybe a million? Sun bug 4756989 was a user complaining that SunOS

Re: getloadavg test

2011-02-17 Thread Bruno Haible
Hi Paul, Sure, something like that sounds fine. Though I'd make the upper bound higher than 100. Maybe a million? ... high load factors are not that uncommon on real-world multiple-CPU machines. Thanks for the advice. I pushed it with 100 as plausibility limit. Bruno -- In memoriam

Re: getloadavg on Solaris 2.6

2011-02-17 Thread Bruno Haible
Hi Paul, When compiling a testdir for the 'getloadavg' module on Solaris 2.6 - which does not have getloadavg in libc, unlike Solaris 7 -, I get this error: cc -O -g -o test-getloadavg test-getloadavg.o ../gllib/libgnu.a ild: (undefined symbol) kstat_read -- referenced in the text segment of

Re: getloadavg on IRIX

2011-02-17 Thread Bruno Haible
Trying to use a getloadavg testdir on IRIX 6.5, the test is skipped: Skipping test; load average not supported: Not supported SKIP: test-getloadavg The reason is that two things are wrong in the execution of getloadavg: 1) ldav_off is set to the value 542000976 = 0x204E4750 but nm /unix

Re: getloadavg and platforms

2011-02-17 Thread Bruno Haible
I had a hard time trying to understand which code is being used on which platform. Here's a proposed patch for clarification: 2011-02-17 Bruno Haible br...@clisp.org getloadavg: Add comments about platforms. * m4/getloadavg.m4: Add comment. * lib/getloadavg.c:

config.rpath update

2011-02-17 Thread Bruno Haible
It's time to update config.rpath, to match libtool-2.4. This update introduces support for GNU/kOpenSolaris, Haibku, tpf, and AmigaOS/PowerPC. 2011-02-17 Bruno Haible br...@clisp.org havelib: Update config.rpath. * build-aux/config.rpath: Update to match libtool.m4 from

new stable snapshot

2011-02-17 Thread Ian Beckwith
Hi, I've released a new stable snapshot. See attached NEWS.stable for details. Tarball: http://erislabs.net/ianb/projects/gnulib/gnulib-20110216-stable.tar.gz Gitweb: http://erislabs.net/gitweb?p=gnulib.git;a=shortlog;h=refs/heads/stable Git: git://erislabs.net/gnulib.git tag: stable/20110216

Re: new module 'setlocale'

2011-02-17 Thread Eric Blake
On 02/12/2011 10:21 AM, Bruno Haible wrote: I'm therefore taking the code from gettext. It provides a workaround against the problems 1 and 3. And here's the fix for problem 2. 2011-02-11 Bruno Haible br...@clisp.org setlocale: Workaround native Windows bug. *

Re: new module 'setlocale'

2011-02-17 Thread Bruno Haible
Hi Eric, I'm now seeing this failure on Haiku alpha 2: FAIL: test-setlocale2.sh (exit: 1) == setlocale did not fail for implicit ar_SA.ISO-8859-1 What can I do to help you diagnose the root cause and work around this issue? It should be easy to trim

Re: [PATCH] vasnprintf: silence some clang false positives

2011-02-17 Thread Bruno Haible
Hi Paul, Eric, Jim, Ben, Paul Eggert wrote: It doesn't feel right that sa_assert uses assert. They should be more independent. I agree with that. I think part of the problem here is the naming convention. Ordinary C assert (X) means crash if X is false. But sa_assert (X) means assume that

Re: getloadavg on Solaris 2.6

2011-02-17 Thread Paul Eggert
On 02/17/2011 04:13 PM, Bruno Haible wrote: getloadavg: Fix link error on Solaris 2.6. Yes, thanks, that looks good.

Re: getloadavg and platforms

2011-02-17 Thread Paul Eggert
On 02/17/2011 04:26 PM, Bruno Haible wrote: getloadavg: Add comments about platforms. Thanks, that looks good too.

Re: new module 'setlocale'

2011-02-17 Thread Eric Blake
On 02/17/2011 07:40 PM, Bruno Haible wrote: Hi Eric, I'm now seeing this failure on Haiku alpha 2: FAIL: test-setlocale2.sh (exit: 1) == setlocale did not fail for implicit ar_SA.ISO-8859-1 What can I do to help you diagnose the root cause and work around

Re: getloadavg on IRIX

2011-02-17 Thread Paul Eggert
On 02/17/2011 04:22 PM, Bruno Haible wrote: I think ldav_off should be changed to 'ptrdiff_t'. Thanks for catching that. I pushed this: * lib/getloadavg.c (getloadavg) [sgi]: Make ldav_off of type ptrdiff_t. It was 'int', but this doesn't match the IRIX 6.5 manual. Suggested by Bruno

Re: new module 'setlocale'

2011-02-17 Thread Bruno Haible
Eric Blake wrote: #include locale.h #include stdio.h #include stdlib.h #include string.h int main () {   if (setlocale (LC_ALL, ar_SA.ISO-8859-1) != NULL)     {       printf (%s\n, setlocale (LC_ALL, NULL));       printf (%s\n, setlocale (LC_CTYPE, NULL));     }  

* NEWS: Mention 2011-02-08 change to stdlib.

2011-02-17 Thread Paul Eggert
I pushed this NEWS entry, which I forgot to do on Feb 8. * NEWS: Mention 2011-02-08 change to stdlib. diff --git a/NEWS b/NEWS index 795488a..b203b2c 100644 --- a/NEWS +++ b/NEWS @@ -23,6 +23,11 @@ DateModules Changes KMEM_GROUP and NEED_SETGID are no

Re: [PATCH] vasnprintf: silence some clang false positives

2011-02-17 Thread Paul Eggert
Ordinary C assert (X) means crash if X is false. But sa_assert (X) means assume that X is true. These are two very different different things. ... for both macros, in the thinking of the programmer, it's an assertion: the programmer asserts (that is, guarantees) a certain condition. The

removing stdint's dependency on wchar

2011-02-17 Thread Paul Eggert
I looked into importing the md5 module into Emacs, and found that it dragged in a whole bunch of stuff related to wchar_t that were irrelevant to md5. The problem is that md5 includes stdint.h for uint32_t, and the stdint module depends on wchar for WCHAR_MIN etc; but md5 does not need WCHAR_MIN