From: Tomi Ollila <tomi.ollila@iki.fi>
To: Jinwoo Lee <jinwoo68@gmail.com>, notmuch@notmuchmail.org
Subject: Re: [PATCH] lib, ruby: make use of -Wl,--no-undefined configurable
In-Reply-To: <87vbem26qk.fsf@maritornes.cs.unb.ca>
References: <1433142542-2307-3-git-send-email-david@tethera.net>
 <1434261524-14139-1-git-send-email-david@tethera.net>
 <yq65a8w2u7yv.fsf@jinwoo-macbookair.roam.corp.google.com>
 <87r3peye4q.fsf@maritornes.cs.unb.ca>
 <yq657fr6tj64.fsf@jinwoo-macbookair.roam.corp.google.com>
 <87oaki3zrh.fsf@maritornes.cs.unb.ca>
 <yq654mm9tyie.fsf@jinwoo-macbookair.roam.corp.google.com>
 <m2bnghe3ar.fsf@guru.guru-group.fi>
 <yq65wpz463rd.fsf@jinwoo-macbookair.roam.corp.google.com>
 <m2lhfjwm4x.fsf@guru.guru-group.fi>
 <yq65y4jjjk2h.fsf@jinwoo-macbookair.roam.corp.google.com>
 <87vbem26qk.fsf@maritornes.cs.unb.ca>
User-Agent: Notmuch/0.20+12~gf3d9440 (http://notmuchmail.org) Emacs/24.3.1
 (x86_64-unknown-linux-gnu)
Date: Wed, 17 Jun 2015 14:16:03 +0300
Message-ID: <m21thad1qk.fsf@guru.guru-group.fi>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: list

On Wed, Jun 17 2015, David Bremner <david@tethera.net> wrote:

> Jinwoo Lee <jinwoo68@gmail.com> writes:
>>  FAIL   PATH is set to build directory
>> 	--- T000-basic.11.expected	2015-06-16 23:40:30.000000000 +0000
>> 	+++ T000-basic.11.output	2015-06-16 23:40:30.000000000 +0000
>> 	@@ -1 +1 @@
>> 	-/Users/jinwoo/Box Sync/personal/project/notmuch
>> 	+
>> usage: dirname path
>
> This failure is because of spaces in your pathname. It might be simpler
> to run in a different directory rather than fix all the problems this
> causes in the test suite.

Yes.. space in pathname also causes that 'ambiquous redirect'. 

Please use a directory which absolute path does not contain spaces (nor
tabs nor newline ;/) -- it would be quite a big project to make test
suite work with whitespace in filename and still something might broke
so it is not worth an effort.

Probably not all tests will succeed anyway, but if you get relevant
tests to work then that should be good enough...

Tomi

>
>>  FAIL   notmuch is compiled with debugging symbols
>> 	--- T000-basic.12.expected	2015-06-16 23:40:30.000000000 +0000
>> 	+++ T000-basic.12.output	2015-06-16 23:40:30.000000000 +0000
>> 	@@ -1 +1 @@
>> 	-1
>> 	+0
>> ./T000-basic.sh: line 95: readelf: command not found
>
> This test, as it says, is testing if notmuch was compiled with debugging
> symbols.  Rather than yet another special case for mac, I'd be curious
> if there is a portable test for this (perhaps using nm?). In any case,
> this one test failing should not be fatal.

yes, there is just one (1) test in the whole suite (last one in T000-basic)
that uses readelf. 

>
>> ./test-lib.sh: line 1062: $test_results_path: ambiguous redirect
>> ./test-lib.sh: line 1063: $test_results_path: ambiguous redirect
>> ./test-lib.sh: line 1064: $test_results_path: ambiguous redirect
>> ./test-lib.sh: line 1065: $test_results_path: ambiguous redirect
>> ./test-lib.sh: line 1066: $test_results_path: ambiguous redirect
>> ./test-lib.sh: line 1067: $test_results_path: ambiguous redirect
>
> At a guess, this again has to do with spaces in the pathname. Maybe this
> is fixable with more quoting, but it's really independent of mac vs
> gnu/linux. In this instance it looks like quoting "test_results_path"
> would help.