Re: gmake-4.4.1 has a performance regression: at least the science/nwchem project now builds much slower
- Reply: Tatsuki Makino : "Re: gmake-4.4.1 has a performance regression: at least the science/nwchem project now builds much slower"
- In reply to: Christian Weisgerber : "Re: gmake-4.4.1 has a performance regression: at least the science/nwchem project now builds much slower"
- Go to: [ bottom of page ] [ top of archives ] [ this month ]
Date: Mon, 08 Apr 2024 14:59:27 UTC
On Mon, 8 Apr 2024 14:29:50 +0200 Christian Weisgerber <naddy@mips.inka.de> wrote: > Tomoaki AOKI: > > > Possibly I'm overlooking something in Porter's Handbook, but is there a > > proper and generic way that builds/installs some build dependencies > > under ${WRKDIR} and use it, without installing into the bare-metal > > environment? > > Yes, you can use BUILD_DEPENDS=${NONEXISTENT}:foo/bar:stage. > > See net/relayd or security/dsniff. > > -- > Christian "naddy" Weisgerber naddy@mips.inka.de Thanks! Is it described somewhere in the Porter's Handbook? If yes, clearly I've overlooked. If no, it should be described somewhere within Chapter 4-6, 12 or 13 in the Porter's Handbook. # Why I've included Chapter 12 "Security"? It's because avoiding any of # BUILD_DEPENDS would be usually the deppendency is known to be vulnerable # and so not desireble to be installed into bare-metal environment. but # still mandatory to build the port. Such BUILD_DEPENDS should be marked # as CONFLICT_INSTALL to all ports. This use case should be benefical regardless # its fundamental messes on build resources. # Another use case I can imagine is to override OPTIONS to build specific # port alone. Regards. -- Tomoaki AOKI <junchoon@dec.sakura.ne.jp>