Re: Broken interactive rebase text after some UTF-8 characters

[Alban Gruin <alban.gruin@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 2316 bytes --]

Hi Michal,

Le 01/02/2019 à 10:06, Michal Nowak a écrit :
> Johannes,
> 
> On Friday, February 1, 2019 at 8:38 AM, Johannes Schindelin <Johannes.Schindelin@gmx.de> wrote:
>> Hi,
>>
>> On Thu, 31 Jan 2019, Junio C Hamano wrote:
>>
>>> Phillip Wood <phillip.wood@talktalk.net> writes:
>>>
>>>>> Are we misusing C formats?
>>>>
>>>> The C standard and POSIX both say that the * refers to the maximum
>>>> number of bytes to print but it looks like it is being treated as the
>>>> maximum number of characters on OpenIndiana.
>>>>
>>>> Johannes - Perhaps we should change it to use fwrite() unless
>>> printf()
>>>> gets fixed and we're sure no other operating systems are affected?
>>>
>>> Avoid such a rewrite, as "%*.s" that takes (int, char *) are used in
>>> many other places in our codebase, if you can.
>>
>> Yes, this would be painful in particular in cases like
>>
>> 	master:advice.c:101:           fprintf(stderr, _("%shint: %.*s%s\n"),
>>
>> where we want to write more than just a variable-length buffer.
>>
>> I am curious: is libintl (gettext) used on OpenIndiana? I ask because
>> AFAIR fprintf() is overridden in that case, and the bug might be a lot
>> easier to fix if it is in libintl rather than in libc.
> 
> here you can see the full output of the OpenIndiana git build: https://hipster.openindiana.org/logs/oi-userland/latest/git.publish.log.
> 
> From what I see there, libintl was found.
> 
> If you believe this is illumos libc bug, it would be cool if someone created an simple testcase, which I can forward to the illumos developers.
> 

I attached a test case to this email.  You can build it with `gcc
test-case.c', and run it with `./a.out'.

Output on my Linux system:

    Before setting locale:
    Expected output:
    áaaa
    Actual output:
    áaaa

    After setting locale:
    Expected output:
    áaaa
    Actual output:
    áaaa

Output on an OpenIndiana system:

    Before setting locale:
    Expected output:
    áaaa
    Actual output:
    áaaa

    After setting locale:
    Expected output:
    áaaa
    Actual output:
    áaaaa

> Thanks,
> Michal
> 
>>
>> Of course, it might *still* be a bug in libc by virtue of handing '%.*s'
>> through to libc's implementation.
>>
>> Alban, can you test this with NO_GETTEXT?
>>
>> Thanks,
>> Johannes

Cheers,
Alban


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: test-case.c --]
[-- Type: text/x-csrc; name="test-case.c", Size: 650 bytes --]

/*
 * Test case for OpenIndiana '%.*s' bug
 * Build with `gcc test-case.c'
 * Run with `./a.out'
 */

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include <locale.h>
#include <libintl.h>

static void compare_output(const char *str, int len) {
	puts("Expected output:");
	fwrite(str, len, sizeof(char), stdout);

	puts("\nActual output:");
	printf("%.*s\n", len, str);
}

int main(int argc, char **argv) {
	char buf[] = "áaaaaaa";

	puts("Before setting locale:");
	compare_output(buf, 5);

	setlocale(LC_ALL, "");

	puts("\nAfter setting locale:");
	compare_output(buf, 5);

	return EXIT_SUCCESS;
}