Re: misc/43644 (TESTS_SUBDIR/atf-run interaction is too fragile)

[Julio Merino <jmmv%NetBSD.org@localhost>]

The following reply was made to PR misc/43644; it has been noted by GNATS.

From: Julio Merino <jmmv%NetBSD.org@localhost>
To: Antti Kantee <pooka%iki.fi@localhost>
Cc: gnats-bugs%netbsd.org@localhost
Subject: Re: misc/43644 (TESTS_SUBDIR/atf-run interaction is too fragile)
Date: Wed, 21 Jul 2010 10:38:32 +0100

 On Wed, Jul 21, 2010 at 7:46 AM, Antti Kantee <pooka%iki.fi@localhost> wrote:
 > On Tue Jul 20 2010 at 16:44:50 +0000, jmmv%NetBSD.org@localhost wrote:
 >> Synopsis: TESTS_SUBDIR/atf-run interaction is too fragile
 >>
 >> State-Changed-From-To: open->suspended
 >> State-Changed-By: jmmv%NetBSD.org@localhost
 >> State-Changed-When: Tue, 20 Jul 2010 16:44:49 +0000
 >> State-Changed-Why:
 >> This is actually an ATF bug. =A0Filed ticket #55 upstream.
 >
 > TESTS_SUBDIR is NetBSD-only, which is why I filed this here.
 
 But the atf-run breakage is not.  atf-run should be more resilent to
 missing test programs/directories.  If that was the case, then you
 would have noticed that the "common" test failed, and would have
 corrected the issue later in the Makefile.
 
 > Having TESTS_SUBDIR behave like SUBDIR for non-test subdirs would be
 > nice to avoid confusion/bugs. =A0If that is fixable just by modifying
 > atf-run, ok.
 
 That'd be nice.  Or rather, not have TESTS_SUBDIR *at all* and let
 SUBDIR DTRT in all cases.  But how do you figure it out, from a parent
 Makefile, that a particular subdirectory will have tests?  If we can
 *ensure* that the parent directory gets processed *after* its children
 (which I don't know), then we could easily inspect and see if any of
 the subdirectories processed by the parent makefile have an Atffile
 and, in those cases, register them into the parent Atffile.  Sounds
 fragile though...
 
 --=20
 Julio Merino