On 2019-Dec-09, Tom Lane wrote:
> I wrote:
> > Alvaro Herrera <alvherre@2ndquadrant.com> writes:
> >> 1. change enough of the build system so that pg_encoding_mbcliplen is
> >> available. (Offhand I see no reason why we couldn't move the
> >> function from mbutils.c to wchar.c, but I haven't tried.)
>
> > I'd be in favor of this if it doesn't turn out to require nasty
> > contortions. My gut feeling (like yours, without having looked) is that
> > the incremental amount of code to be moved into wchar.c wouldn't be much.
>
> Actually, on second thought --- the issue here is not so much how much new
> code shows up in libpgcommon, and more that executables using stringinfo.o
> will now find themselves pulling in all of wchar.o, where before they
> might've pulled in none of it. We need to look at how much code bloat
> ensues. In the end it might be smart to put this new functionality in
> some separate source file instead of dropping it into stringinfo.c.
> (It could also be that all the executables that need stringinfo.o are
> already pulling in wchar functionality for other reasons, but we should
> check the code-size implications before assuming this is fine.)
Hmm, that's a good thought. On the one hand, nothing in the frontend
needs this new function ... the only current use is pg_waldump which
likely doesn't need to quote anything, and I just looked at Andres'
programmatically- written node support code which is why we have
StringInfo in frontend in the first place -- it also doesn't care.
So rather than mess with stringinfo.c at all I could just create
stringinfo_server.c and put this function there, compiled only for
backend ...
--
Álvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
