mkdir-p.c race condition fix for setuid/setgid/sticky directories

I installed this: 2007-01-31 Paul Eggert <[EMAIL PROTECTED]> * lib/mkdir-p.c (make_dir_parents): Close a race condition that occurs when "mkdir -m foo" creates a setgid directory that is (1) writeable to group or other and (2) is intended to have a special mode b

Re: enforcing the use of string.h related modules

Eric Blake byu.net> writes: > > I see that while I was typing this, you checked in your original proposed > change. But with your patch applied, I get the following: Another one: gcc -std=gnu99 -I. -I/usr/local/include -g2 -Wall -Werror -MT mountlist.o - MD -MP -MF .deps/mountlist.Tpo -

Re: enforcing the use of string.h related modules

Eric Blake byu.net> writes: > Contrary to the Changelog entry, I don't see gl_STRING_MODULE_INDICATOR defined > in string_h.m4. > I see that while I was typing this, you checked in your original proposed change. But with your patch applied, I get the following: gcc -I. -I../../lib -g2

Re: enforcing the use of string.h related modules

Eric Blake wrote: > Contrary to the Changelog entry, I don't see gl_STRING_MODULE_INDICATOR > defined > in string_h.m4. Oops, forgot to commit this one. Sorry. Should be better now. Bruno

Re: enforcing the use of string.h related modules

Bruno Haible clisp.org> writes: > > > Here is a proposed patch, that leads to a link error if someone > > > uses stpcpy() without the 'stpcpy' module (assuming the 'string' module > > > is present - which is quite likely since so much depends on it). > > > There were also no objections, so I appl

putting external library dependencies into the module description

Hi, Ralf noticed that since the latest change to copy-file.c the gettext CVS does not build any more on some platforms. The problem is the dependency to libacl, which is currently not documented - not in comments of m4/acl.m4, and not in the module descriptions. We also had similar problems in the

Re: enforcing the use of string.h related modules

Eric Blake wrote: > > Here is a proposed patch, that leads to a link error if someone > > uses stpcpy() without the 'stpcpy' module (assuming the 'string' module > > is present - which is quite likely since so much depends on it). > > Sounds like a great idea to me. OK, thanks for this opinion. T

gai_strerror

Hello, There is no message for the EAI_OVERFLOW error in gai_strerror. The getnameinfo can return the EAI_OVERFLOW error. Patch attached. Jean-Louis Index: lib/gai_strerror.c === --- lib/gai_strerror.c (revision 5025) +++ lib/gai_

Re: gai_strerror and EAI_OVERFLOW

Jean-Louis Martineau <[EMAIL PROTECTED]> writes: > The gai_strerror function doesn't return an error message for the > EAI_OVERFLOW error code. > > getnameinfo can return the EAI_OVERFLOW error. Thanks, installed. /Simon

gai_strerror and EAI_OVERFLOW

The gai_strerror function doesn't return an error message for the EAI_OVERFLOW error code. getnameinfo can return the EAI_OVERFLOW error. Patch attached. Jean-Louis Index: lib/gai_strerror.c === --- lib/gai_strerror.c (revision 50