On 06/21/2013 08:32 AM, HATAYAMA Daisuke wrote:
> (2013/06/20 15:43), Atsushi Kumagai wrote:
>> Hello Baoquan and HATAYAMA-san,
>>
>> On Thu, 20 Jun 2013 13:08:17 +0800
>> Baoquan <bhe at redhat.com> wrote:
>>
>>> On 06/20/2013 08:01 AM, HATAYAMA Daisuke wrote:
>>>
>>>>
>>>>> makedumpfile has some options which has two 2 forms, long option
>>>>> and short option. E.g. --help | -h
>>>>>
>>>>> In man page and help, only one of them are presented, user may
>>>>> complain
>>>>> of this. Here add missing options.
>>>>>
>>>>
>>>> I guess the maintainer does so intentionally, I'm not sure of course.
>>>> Characters for short options is limited resource, but now makedumpfile
>>>> has considerably many options. We can no longer assign reasonable
>>>> letters to each functionality with short options. For example, please
>>>> look at the definition of longopts:
>>
>> That's right, some letters assigned to recent feature have no reason.
>> So, they aren't suitable to be exported explicitly for users.
>>
>>>>
>>>> static struct option longopts[] = {
>>>> {"split", no_argument, NULL, 's'},
>>>> {"reassemble", no_argument, NULL, 'r'},
>>>> {"xen-syms", required_argument, NULL, 'y'},
>>>> {"xen-vmcoreinfo", required_argument, NULL, 'z'},
>>>> {"xen_phys_start", required_argument, NULL, 'P'},
>>>> {"message-level", required_argument, NULL, 'm'},
>>>> {"vtop", required_argument, NULL, 'V'},
>>>> {"dump-dmesg", no_argument, NULL, 'M'},
>>>> {"config", required_argument, NULL, 'C'},
>>>> {"help", no_argument, NULL, 'h'},
>>>> {"diskset", required_argument, NULL, 'k'},
>>>> {"non-cyclic", no_argument, NULL, 'Y'},
>>>> {"cyclic-buffer", required_argument, NULL, 'Z'},
>>>> {"eppic", required_argument, NULL, 'S'},
>>>> {"map-size", required_argument, NULL, 'A'},
>>>> {0, 0, 0, 0}
>>>> };
>>>>
>>>> I don't see any relationship on the last three options: non-cyclic
>>>> vs Y,
>>>> cyclic-buffer vs Z, eppic vs S and map-size vs A. I don't think there
>>>> are users who use short options for these features.
>>>
>>>
>>> Hi,
>>>
>>> Thanks for telling. I totally understand that unmarched letters between
>>> long and short options are truly annoying. But like current code, only
>>> presenting long options in man page and help may cause complaint from QA
>>> or customers.
>>>
>>>>
>>>> One of my solution is conversely to delete short options, and I don't
>>>> think it problem because the short options are only implicitly exported
>>>> without explicit description on such as manual page and help message.
>>>
>>>
>>> Yeah, deleting short options is also a good method, let's wait and see
>>> what the maintainer say.
>>
>> I agree to delete short options.
>> Could you re-send the patch ?
>>
>
> BTW, we can assign non-printable characters as short options. Normally
> we don't type non-printable characters on console, so it's for free.
>
Hi,
I think it's good idea. In kexec, some options use below short option
with a large value. With these indicative MACRO as short option, it will
make code more readable.
Then all old short options can be replaced with them.
#define OPT_MEM_MIN 256
#define OPT_MEM_MAX 257
#define OPT_REUSE_INITRD 258
#define OPT_LOAD_PRESERVE_CONTEXT 259
#define OPT_LOAD_JUMP_BACK_HELPER 260
#define OPT_ENTRY 261
#define OPT_MAX 262
Baoquan
Thanks a lot
