Thread: Remove PARTIAL_LINKING?
Hi, We've still some support for building the backend with PARTIAL_LINKING / SUBSYS.o instead of the current objfiles.txt approach. http://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=9956ddc19164b02dc1925fb389a1af77472eba5e Any objections to removing that? Seems that's largely just removing a bit of logic from common.mk. Note I'm not talking about whatever that aix logic in src/backend is, that seems independent except for sharing the SUBSYS.o name. I just encountered this while polishing my JIT patch, which has *optional* support for building inlinable version of backend functions. It'd not be hard to make it work for PARTIAL_LINKING, but I kinda don't see the point 10 years later... Greetings, Andres Freund
Andres Freund <andres@anarazel.de> writes: > We've still some support for building the backend with PARTIAL_LINKING / > SUBSYS.o instead of the current objfiles.txt approach. > Any objections to removing that? +1. I think the reason for holding onto the old code was mainly fear of hitting command-line length limits in the link step. Since that still hasn't happened 10 years later, even on the oldest buildfarm critters, it seems moot. We'd probably just decide to desupport any platform where that did happen in future. regards, tom lane
On 1/21/18 15:43, Andres Freund wrote: > We've still some support for building the backend with PARTIAL_LINKING / > SUBSYS.o instead of the current objfiles.txt approach. > http://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=9956ddc19164b02dc1925fb389a1af77472eba5e > > Any objections to removing that? Seems that's largely just removing a > bit of logic from common.mk. Yeah, I think it's OK to remove that. -- Peter Eisentraut http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
Hi, On 2018-01-21 12:43:56 -0800, Andres Freund wrote: > We've still some support for building the backend with PARTIAL_LINKING / > SUBSYS.o instead of the current objfiles.txt approach. > http://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=9956ddc19164b02dc1925fb389a1af77472eba5e > > Any objections to removing that? Seems that's largely just removing a > bit of logic from common.mk. > > Note I'm not talking about whatever that aix logic in src/backend is, > that seems independent except for sharing the SUBSYS.o name. > > > I just encountered this while polishing my JIT patch, which has > *optional* support for building inlinable version of backend > functions. It'd not be hard to make it work for PARTIAL_LINKING, but I > kinda don't see the point 10 years later... And removed. Note the SUBSYS.o reference from aix appears to be independent... - Andres