This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH v3][BZ 14561] Testcase
- From: "Carlos O'Donell" <carlos at redhat dot com>
- To: OndÅej BÃlka <neleai at seznam dot cz>
- Cc: libc-alpha at sourceware dot org
- Date: Tue, 11 Jun 2013 10:21:32 -0400
- Subject: Re: [PATCH v3][BZ 14561] Testcase
- References: <20130520190628 dot GA22088 at domone dot kolej dot mff dot cuni dot cz> <519E93F3 dot 8060203 at redhat dot com> <20130603204113 dot GA26974 at domone dot kolej dot mff dot cuni dot cz> <51ADE6C1 dot 1030806 at redhat dot com> <20130604142546 dot GA32680 at domone dot kolej dot mff dot cuni dot cz> <51ADFD4C dot 5000807 at redhat dot com> <20130611090519 dot GB6224 at domone dot kolej dot mff dot cuni dot cz>
On 06/11/2013 05:05 AM, OndÅej BÃlka wrote:
>> Mention BZ#.
>>
> Done.
Thanks.
>>> +do_test (void)
>>> +{
>>> + char buf[1000], *p;
>>
>> Big enough?
>>
> Buffer here is used only for single sequence and then reset to
> beginning. A 1000 bytes should be enough.
>
> I was ending on first failure so I decided to collect all failures.
I also like that. Terminating on first failure is problematic because
it means you need to keep re-running the tests to detect subsequent
failures and that's annoying. Thanks.
>>> +#define FAIL { \
>>> + fputs ("Inconsisted results for sequence:\n",stderr); \
>>> + fputs (buf, stderr); \
>>> + return 1; \
>>> +}
>>
>> Please make this a function so you can easily put a breakpoint on it.
>>
> Done.
Thanks.
>> Looks good to me with above two changes and proof
>> that 1000 bytes is enough, otherwise use 8kb, which
>> is certainly large enough for 6144 (64*6*13) bytes
>> of messages.
>>
> When this is good and rand is also good is this OK to commit them both?
This test looks good to me now.
What is the status of the rand changes? Did Roland request a new symbol
or are we OK with just changing this behaviour?
> Ondra
>
> * bug-srand-srandom-dependency.c: New file.
> * Makefile (tests): Add it.
OK for me.
Cheers,
Carlos.