Re: pg_dump.options.diff

Поиск
Список
Период
Сортировка
Искать
От
Tom Lane
Тема
Re: pg_dump.options.diff
Дата
Msg-id
4981.1041549367@sss.pgh.pa.us
Ответ на
Re: pg_dump.options.diff (Serguei Mokhov)
Список
Дерево обсуждения
Re: pg_dump.options.diff "Serguei Mokhov" <mokhov@cs.concordia.ca>
Re: pg_dump.options.diff Tom Lane <tgl@sss.pgh.pa.us>
Re: pg_dump.options.diff Peter Eisentraut <peter_e@gmx.net>
Re: pg_dump.options.diff Tom Lane <tgl@sss.pgh.pa.us>
Re: pg_dump.options.diff Peter Eisentraut <peter_e@gmx.net>
"Serguei Mokhov"  writes:
>> But there's no good place to put it.  I'd say just stick it into each
>> tool; it's no worse than repeating the existence of a "usage()"
>> subroutine in each tool.

> It ended up being in dumputils.h

I really don't like putting a macro with a name as short as "xo" into a
header file, even one of relatively narrow scope.  It's too likely to
create weird conflicts.  My inclination is to make the coding be more
like

static void
usage(void)
{
#if defined(HAVE_GETOPT_LONG)
#define xo(longOption,shortOption,desc)  printf("%s %s\n", longOption, desc)
#else
#define xo(longOption,shortOption,desc)  printf("%s %s\n", shortOption, desc)
#endif

	... lots of xo() calls ...

#undef xo
}

This gives us the convenience of a very short name within the usage()
subroutines, while not polluting the namespace for everyplace else in
these utilities.

As I said before, duplicating the definition of xo() in each file that
uses it doesn't bother me a bit; it's too simple for that to be a
significant objection.

			regards, tom lane
В списке pgsql-hackers по дате отправления
От: Peter Eisentraut
Дата:
От: Lamar Owen
Дата:
Сообщение: Upgrading rant.
FAQ