> What is the reason? Can malloc() return NULL with errno being set to something
> different than ENOMEM?
It is probably just overly-conservative code in those callers.
Eric Blake wrote:
> Here's what I'm committing; I think it addresses all your points.
>
> 2007-08-22 Eric Blake <[EMAIL PROTECTED]>
>
> Getline touchups.
> * lib/getdelim.c (getdelim): Revert regression that required *n to
> be 0 when *lineptr is NULL. Preserve errno across f
Jim Meyering wrote:
> You've expanded gnulib's set of target systems to include some with
> non-conforming malloc functions.
Coreutils does indeed not attempt portability to mingw.
However, gettext and libiconv do support mingw since 2006; other packages
relying on gnulib maybe did so earlier tha
> At these places malloc/realloc/calloc NULL results are not well handled, IIUC.
I fixed the hash.c one:
http://git.sv.gnu.org/gitweb/?p=gnulib.git;a=summary
* lib/hash.c (hash_initialize): Detect calloc failure.
Reported by Bruno Haible.
Bruno Haible <[EMAIL PROTECTED]> wrote:
> Hi,
>
> At these places malloc/realloc/calloc NULL results are not well handled, IIUC.
>
> Sergey:
> argp-help.c:440
> argp-help.c:463
> argp-help.c:464
> argp-help.c:877
> argp-help.c:880
> Simon:
> gc-gnulib.c:583
> Jim:
> hash.c:578
Good
Bruno Haible <[EMAIL PROTECTED]> wrote:
> I found these missing dependencies. If you disagree to these dependencies,
> you are free to remove them and set errno to ENOMEM manually in the .c file.
>
> 2007-09-09 Bruno Haible <[EMAIL PROTECTED]>
>
> * modules/getcwd (Depends-on): Add malloc-p
The 'canonicalize-lgpl' module should better not rely on malloc-posix or
realloc-posix, otherwise it adds extra complexity to the 'relocatable'
module family. Setting errno = ENOMEM is the more maintainable solution here.
2007-09-09 Bruno Haible <[EMAIL PROTECTED]>
* lib/canonicalize-l
Hi,
I found these missing dependencies. If you disagree to these dependencies,
you are free to remove them and set errno to ENOMEM manually in the .c file.
2007-09-09 Bruno Haible <[EMAIL PROTECTED]>
* modules/getcwd (Depends-on): Add malloc-posix.
* modules/glob (Depends-on):
Hi,
At these places malloc/realloc/calloc NULL results are not well handled, IIUC.
Sergey:
argp-help.c:440
argp-help.c:463
argp-help.c:464
argp-help.c:877
argp-help.c:880
Simon:
gc-gnulib.c:583
Jim:
hash.c:578
Paul:
regcomp.c:864
regcomp.c:3030 missing cleanup action, leaks memo
Jim Meyering wrote:
> POSIX guarantees that errno is defined when malloc or realloc
> fails, so there's no need to set it manually in that case.
Yet glibc does set it manually in some cases:
glibc-2.6.1/login/openpty.c:69
glibc-2.6.1/login/login.c:67
glibc-2.6.1/stdlib/setenv.c:180
glibc-2.6.1/sy
Jim Meyering wrote:
> > So I propose to add three modules 'malloc-posix', 'realloc-posix',
> > 'calloc-posix' (using the '-posix' suffix that we already know from
> > 'fnmatch-posix'/'fnmatch-gnu' and 'printf-posix').
>
> I'm glad you've seen the light! :-)
I take this as an OK to the proposal, s
Jim Meyering wrote on 2007-09-04:
> How about adding a way to tell gnulib-tool
> that a module (or just one of its names) is deprecated. Then, assuming
> both malloc and malloc-gnu module names are accepted and equivalent
> for several months -- while gnulib-tool emits warnings[*] -- remove the
>
Jim Meyering wrote on 2007-09-04:
> Currently, gnulib-tool emits so much output that
> a few new diagnostics may easily go by unnoticed.
This has been mentioned several times lately. I'm adding an option --quiet
that decreases the verbosity and an option --verbose that cancels --quiet
(and vice ve
13 matches
Mail list logo
