Developers of Sympa

[Guillaume Rousse <address@concealed>]

Le 06/06/2014 09:10, IKEDA Soji a écrit :
> Hi, please tell me...
>
> On Wed,  4 Jun 2014 17:28:26 +0200 (CEST)
> address@concealed wrote:
>
> > sympa[10828] trunk/t/Language.t: [dev] no need to re-add standard @INC, but 
> > relative path for src/lib helps executing the tests manually
> > Revision 10828 Author rousse Date 2014-06-04 17:28:25 +0200 (mer. 04 juin 
> > 2014)
> > Log Message[dev] no need to re-add standard @INC, but relative path for 
> > src/lib helps executing the tests manually
> > Modified Paths
> > trunk/t/Language.t
> > Diff
> > Modified: trunk/t/Language.t (10827 => 10828)
> > --- trunk/t/Language.t  2014-06-04 13:14:25 UTC (rev 10827)
> > +++ trunk/t/Language.t  2014-06-04 15:28:25 UTC (rev 10828)
> > @@ -9,9 +9,12 @@
> >
> >   use strict;
> >   use warnings;
> > +
> > +use FindBin qw($Bin);
> > +use lib 't/stub', "$Bin/../src/lib";
> > +
> >   use Test::More;
> >
> > -use lib 't/stub', @INC;
> >   use Sympa::Language;
> >
> >   # Lang 2 gettext locale
>
> @INC was unneccessary.  However, I have two questions:
>
> - Is arrangement by FindBin required?
>
>    Is it possible that "make check" invokes test scripts from the
>    other location than the top of source?  If it is not, lib list
>    can simply contain "src/lib".
make check, no. Direct invocation, yes.

> - May each test script be executable? I mean that its executable
>    bit is on and it has sh-bang line.
>
>    Sh-bang line fixes path of Perl interpreter.  Developer may wish
>    to test with other Perl executables.  I guess the script
>    itself should not have sh-bang and may not be executable.
If you want to use another interpreter, just call it explicitely, and 
pass it the test script as an argument. Neither the sh-bang, nor the 
executable bit prevent you to do it

> Additionally, if "make check" gives default module search path
> using PERL5LIB enviroment variable, lib list also need not contain
> "src/lib".
>
> # n.b. environment variable seems better than -I option,
> # because it will propagate to child processes (see LockedFile.t).
Indeed, altough -I flag can also be passed to child process.

Those scripts are currenly self-contained, executable without 
configuring environment first, while not preventing any override. For 
the cost of two additional line of code... Why should we remove them, 
just because there is another way to do it ?
--
Guillaume Rousse
INRIA, Direction des systèmes d'information
Domaine de Voluceau
Rocquencourt - BP 105
78153 Le Chesnay
Tel: 01 39 63 58 31

Attachment: smime.p7s
Description: Signature cryptographique S/MIME