Re: remove ancient pre-dlopen dynloader code - Mailing list pgsql-hackers

From Andres Freund
Subject Re: remove ancient pre-dlopen dynloader code
Date
Msg-id 20180816141005.nkw5nwkmoip57ui6@alap3.anarazel.de
Whole thread Raw
In response to Re: remove ancient pre-dlopen dynloader code  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: remove ancient pre-dlopen dynloader code
List pgsql-hackers
On 2018-08-16 10:07:04 -0400, Tom Lane wrote:
> Andres Freund <andres@anarazel.de> writes:
> > On 2018-08-16 09:22:14 -0400, Tom Lane wrote:
> >> The lazy man's way to get rid of it would be to put something like
> >> "int bogus = 0;" in the empty dynloader.c files.  Better would be
> >> to not have the empty .c files at all, but I'm not sure how much
> >> we'd have to contort the Makefiles to support that.
> 
> > If I had my druthers, we'd just remove all that configure magic for
> > selecting these files and just use ifdefs.  Personally I find it
> > occasionally that they're linked into place, rather than built under
> > their original name.
> 
> Even if we all agreed that was an improvement (which I'm not sure of),
> it wouldn't fix this problem would it?  On affected platforms, the
> file would still be empty after preprocessing.

Well, that depends on what you put into that file, it seems
realistically combinable with a bunch of non-conditional code...

Anyway, I'm not planning to do something here right now besides putting
-Wno-empty-translation-unit into my scripts, I just wanted to make sure
people are aware that we hit this now.

Greetings,

Andres Freund


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: remove ancient pre-dlopen dynloader code
Next
From: Andres Freund
Date:
Subject: Re: C99 compliance for src/port/snprintf.c