This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [2.20] [2/6] Generate .test-result files for tests with special rules
- From: "Joseph S. Myers" <joseph at codesourcery dot com>
- To: Brooks Moses <bmoses at google dot com>
- Cc: <libc-alpha at sourceware dot org>
- Date: Wed, 12 Feb 2014 22:57:07 +0000
- Subject: Re: [2.20] [2/6] Generate .test-result files for tests with special rules
- Authentication-results: sourceware.org; auth=none
- References: <Pine dot LNX dot 4 dot 64 dot 1401100208000 dot 9412 at digraph dot polyomino dot org dot uk> <Pine dot LNX dot 4 dot 64 dot 1401100211060 dot 9412 at digraph dot polyomino dot org dot uk> <52F953A1 dot 1050306 at google dot com>
On Mon, 10 Feb 2014, Brooks Moses wrote:
> > * Should all the tests generating output files not named
> > <something>.out be changed (in a separate patch) to use the .out
> > naming convention?
>
> IMO, yes. Conventions like this are a useful step towards a system that can
> download them from a remote target, when we eventually get there.
I have a followup question here:
Suppose we make all tests use <something>.out. Then for most purposes we
don't actually need the tests-special makefile variable at all; the tests
could just go in the tests variable, and the makefile rules for the
specific .out files would override the pattern rule. So do we actually
want the tests-special variable at all for cases where the test generates
a .out file, or only for the cases where there is some good reason for the
test to do otherwise (e.g. in catgets/Makefile where the test is that a
.cat file is generated without error, or if we generally say that any
intermediate rule running something on the host counts as a test for the
purpose of allowing the testsuite to run to completion)?
--
Joseph S. Myers
joseph@codesourcery.com