Re: [PATCH] Use O_BINARY to detect whether to pass "rb" to popen

Filipe Brandenburger <[email protected]> Fri, 22 May 2015 08:43:39 -0700
Newsgroups gmane.comp.gnu.utils.bugs
Message-ID <CADU+-uCY4eucZrJ78qNHRu=wzsDr9BqS=RVYtcO7+t6e_tSwOw@mail.gmail.com>
Hi Bruce,

On Fri, May 22, 2015 at 8:32 AM, Bruce Korb <[email protected]> wrote:
> On 05/22/15 02:15, Andreas Schwab wrote:
>>>     {
>>> -    FILE * in_fp = popen (cmdline, "r");
>>> +    FILE * in_fp = popen (cmdline, "r" TEXT_MODE);
>>
>> Didn't you mean FOPEN_TEXT?
>
> Indeed, and yet it built and went through "make distcheck".

That code was under a !HAVE_WORKING_FORK ifdef so it wasn't being
built... I reproduced the build failure on Linux with ./configure
ac_cv_func_fork_works=no instead.

I also think we should get rid of FOPEN_{READ,WRITE}_BINARY and put
these defines in lib/system.h instead of src/local.h.git.

I'll send you an updated patch with that.

> And, yes, the patch included a non-related issue.  Someone came up
> with a tortured uu-encoded file that would cause uudecode to seg fault.
> The patch was against the v4.15.1 tag.

Yeah in that case they should probably be separate patches. I'll send
you a sequence you can grab with "git am". I'm almost done testing
them, so expect them shortly.

Cheers,
Filipe