[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [lmi] [lmi-commits] [6055] Mark lines needing '--gui_test_path'
From: |
Vadim Zeitlin |
Subject: |
Re: [lmi] [lmi-commits] [6055] Mark lines needing '--gui_test_path' |
Date: |
Wed, 10 Dec 2014 14:46:28 +0100 |
On Wed, 10 Dec 2014 12:36:13 +0000 Greg Chicares <address@hidden> wrote:
GC> Revision: 6055
GC> http://svn.sv.gnu.org/viewvc/?view=rev&root=lmi&revision=6055
GC> Author: chicares
GC> Date: 2014-12-10 12:36:11 +0000 (Wed, 10 Dec 2014)
GC> Log Message:
GC> -----------
GC> Mark lines needing '--gui_test_path'
Hello,
I have a question about this patch as it goes against what I was doing to
implement support for --gui_test_path so far and I wonder if this is
intentional. To be precise, this patch seems to strongly hint that you'd
like to modify the test code to use something different from
configurable_settings::instance().default_input_filename() to obtain the
file name to use in the test. My plan was however to keep the test code
here unchanged and instead ensure that default_input_filename() returns
what we want it to return, i.e. uses the value of --gui_test_path command
line option, if any.
I thought of doing it like this because we need to make a backup of
configurable_settings.xml file anyhow, to prevent it from being modified by
the tests (notably the calculation summary one), and we could also adjust
the values to it while doing this. It's not completely trivial because the
class configurable_settings doesn't provide any way of modifying the
default input filename field, so it would have to be done entirely outside
of it, but it seemed to me that it was still worth doing it because we
could also need to modify other configurable settings for the test purposes
(e.g. spreadsheet file extension comes to mind). Am I overcomplicating
things needlessly here or is there some merit to this line of thought?
Also, I assume it's intentional that no comments about using
--gui_test_path were added to the other occurrences of
configurable_settings::instance().default_input_filename() in the testing
code, i.e. those in configurable_settings, default_update, and
input_validation tests. And I suspect this is because these tests are only
meant to be run when --distribution option is given. And this, in turn,
seems to imply that --distribution and --gui_test_path are incompatible. Is
this correct? I.e. should the test give an error if both of them are
specified?
Thanks in advance,
VZ
- Re: [lmi] [lmi-commits] [6055] Mark lines needing '--gui_test_path',
Vadim Zeitlin <=