Re: [PATCH] test-fclose: avoid over-specification (leading to abort)

[Jim Meyering]

Eric Blake wrote:
> On 05/12/2011 08:55 AM, Eric Blake wrote:
>> On 05/12/2011 08:52 AM, Jim Meyering wrote:
>>> Hi Eric,
>>>
>>> This test would fail for me when I made iwhd use the latest from gnulib:
>>>
>>>     FAIL: test-fclose (exit: 134)
>>>     =============================
>>>     test-fclose.c:77: assertion failed
>>>
>>> It looks pretty obvious, but I'll wait for your ACK, just in case.
>>
>> NACK.
>>
>> The whole point of this test is to expose the glibc bug.
>>
>> fclose() is _required_ to rewind seekable input streams back to the
>> point of the last byte returned by fgetc().  That is, even though
>> fgetc() advanced the position to EOF, the fclose() must undo that advance.
>>
>> More likely, this means all the recent flurry in lib/fclose.c has
>> somehow introduced a regression :(
>
> Then again, './gnulib-tool --with-tests --test fclose' worked for me.
> Is this a case of you building incrementally but the replacement
> <stdio.h> for iwhd hasn't been replaced to reflect the change in
> configure tests for fclose?  Could it be a case of missing dependencies
> where m4 macros aren't being expanded early enough?

Argh.  That's it, exactly.
Sorry for the false alert.