[PATCH] Fix install-tests target for vpath builds - Mailing list pgsql-hackers

From Craig Ringer
Subject [PATCH] Fix install-tests target for vpath builds
Date
Msg-id CAMsr+YGCZkDzxRNacKWC9vzJfw54k0CxKxA-d9fEdiZvLQPjHQ@mail.gmail.com
Whole thread Raw
Responses Re: [PATCH] Fix install-tests target for vpath builds  (Craig Ringer <craig@2ndquadrant.com>)
Re: [PATCH] Fix install-tests target for vpath builds  (Andrew Dunstan <andrew.dunstan@2ndquadrant.com>)
List pgsql-hackers
Patch 0001 fixes this issue with vpath postgres builds:

$ make -C src/test/regress install-tests
/usr/bin/install: cannot create regular file 'PGPREFIX/lib/postgresql/regress/PGPREFIX/src/test/regress/expected/errors.out': No such file or directory
make: *** [GNUmakefile:90: install-tests] Error 1

(where PGPREFIX is your --prefix)

It also makes the install-tests target a toplevel target for convenience.

Three related bonus patches are attached in case anyone thinks they're a good idea:

- 0002 changes the install location of src/test/regress's install-tests output files (sql/, expected/ etc) to $(pkglibdir)/pgxs/src/test/regress so that PGXS resolves it as $(top_srcdir)/src/test/regress, same as for in-tree builds. Presently it installs in $(pkglibdir)/regress/ for some reason. This patch applies on top of 0001. It will affect packagers.

- 0003 makes the toplevel install-tests target also install src/test/isolation test resources and the test modules. This patch applies on top of either 0001 or 0002, depending on whether you want to include 0002.

- 0004 makes the dummy 'check' target in pgxs.mk optional for extensions that define the new PGXS variable NO_DUMMY_CHECK_TARGET . This lets extensions that want to define a 'check' target do so without having make complain at them about redefined targets. This patch is independent of the others and can apply on master directly.



--
 Craig Ringer                   http://www.2ndQuadrant.com/
 2ndQuadrant - PostgreSQL Solutions for the Enterprise
Attachment

pgsql-hackers by date:

Previous
From: David Rowley
Date:
Subject: Re: Parallel Append can break run-time partition pruning
Next
From: Craig Ringer
Date:
Subject: Re: [PATCH] Skip llvm bytecode generation if LLVM is missing