On Wed, Jan 2, 2013 at 5:36 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Magnus Hagander <magnus(at)hagander(dot)net> writes:
>>> On Wed, Jan 2, 2013 at 4:14 AM, Andrew Dunstan <andrew(at)dunslane(dot)net> wrote:
>>>> This seems to have broken plperl builds on Windows.
>> I'm not really sure what the best thing is to do here. We can either
>> use the fact that we know that uid_t and gid_t are "int" on all our
>> platforms, more or less, and change the tar api to use uid_t. Or put
>> the tar functions in their own header and make sure we don't include
>> that one anywhere that we include the perl headers.
> Why are these very tar-specific functions being declared in such a
> globally visible spot as port.h? That seems like a bad idea on its
> face. IMO stuff in port.h ought to be about as globally applicable
> as, say, malloc().
It's where we put most of the things from src/port in.
I take it you suggest moving it to a special say include/pgtar.h file?
> If there's actually a reason for that sort of namespace abuse, then
> change their parameters to int --- IIRC, the tar format can't support
> UIDs wider than 32 bits anyway.
Andrew suggested #ifndef'ing the declarations the same way that the
typedefs for uid_t and gid_t are done as another optin. I don't really
find either one of them non-kludgy :)
In response to
pgsql-hackers by date
|Next:||From: Boszormenyi Zoltan||Date: 2013-01-02 16:44:12|
|Subject: Re: Big disconnect_and_exit cleanup in pg_basebackup|
|Previous:||From: Boszormenyi Zoltan||Date: 2013-01-02 16:36:23|
|Subject: Re: allowing multiple PQclear() calls|
pgsql-committers by date
|Next:||From: Andrew Dunstan||Date: 2013-01-02 16:43:14|
|Subject: Re: pgsql: Fix background workers for EXEC_BACKEND|
|Previous:||From: Tom Lane||Date: 2013-01-02 16:36:01|
|Subject: Re: pgsql: Unify some tar functionality across different parts|