Re: [lmi] Manual test of automated GUI test

[Greg Chicares]

On 2014-10-19 00:32Z, Vadim Zeitlin wrote:
> On Sun, 19 Oct 2014 00:02:50 +0000 Greg Chicares <address@hidden> wrote:
[...]
> GC> 22:45:17: Error: can't open file 'wx_test.conf' (error 2: the system 
> cannot find the file specified.)
> GC> 22:45:17:     benchmark_census: ERROR (Assertion failure: can't read from 
> closed file [file ../src/common/ffile.cpp, line 128, in Read()].)
> 
>  These two errors are redundant but I'm not sure if it's worth improving
> this as this file is supposed to exist in the normal case. Do you think it is?

I agree: no extra work here would be worthwhile.

[...skipping self-explanatory items...]

> GC> 22:45:28:     paste_census: ok (1017ms)
> GC> 22:45:36:     pdf_census: ok (8717ms)
> GC> In file ../src/msw/mimetype.cpp at line 272: 'AssocQueryString' failed 
> with error 0x80070002 (the system cannot find the file specified.).
> 
>  This probably indicates that you don't have a PDF viewer (i.e. a program
> registered for opening files with .pdf extension) installed?

But I do have such a thing installed--I won't advertise proprietary
software on a gnu.org mailing list, but it's the first one that
would come to mind in the msw world. And it works with lmi--e.g.,
if I start lmi and do
  File | New | Illustration | OK
  File | Print preview
then "***** Reader" opens and displays 'unnamed1.pdf' correctly.

'emit_ledger.cpp' does it this way:
        file_command()(pdf_out_file, "open");

I randomly guessed that the test might be using a different "verb",
but it doesn't do any mime stuff directly--it just simulates the
menu action that I use successfully with the production system:

    // Launch the PDF creation as side effect of previewing it.
    wxUIActionSimulator ui;
    ui.Char('v', wxMOD_CONTROL);    // "File|Print preview"

so I'm utterly puzzled by this.

> GC> There must be a 'configurable_settings.xml' file somewhere, but
> GC> /etc/opt/lmi/ = "C:\etc\opt\lmi" is only the first location inspected.
> 
>  Yes, I'm aware about this, but I thought the point of this test was to
> check for its existence at this particular location. Especially as it was
> the very first item of the test specification, quoting Wendy's email from
> 2013-11-13T14:51:30Z:
> 
>>      [Specifications begin]
>>
>>      1. Inspect 'configurable_settings.xml' for the following:
>>
>>        A. The file exists in '/etc/opt/lmi'

Here, I'll have to defer to Wendy or Kim. This may be an actual
requirement for assembling a distribution; I don't know.

> GC> I normally don't have a copy in /etc (as noted in this makefile comment):
> 
>  To be honest, neither did I, before I started working on this. But I had
> to create a copy there just to allow this test to pass locally. I could,
> and would probably be glad to, to relax this test and just check that the
> configurable settings file is available somewhere where it can be found.
> But, to be honest, I don't really see the point of this test then as the
> absence of configurable_settings.xml would result in a fatal error anyhow,
> whether we test for it explicitly or not.

Yes, exactly.

>  To summarize, I'm not sure what am I supposed to do here? Modify the test?
> Remove it entirely? Something else?

I'm not the one who can answer this.

>  I'd just like to remind you that you can exclude this test if you don't
> want to run it by using "-t -configurable_settings" command line option.

Yes, thanks for designing it that way--I'm sure it will be useful to me.

[...tests for 'expiry'...]
> Here I'm almost certain that the checks should not be relaxed and IMHO
> using "-t -expiry_dates" would be a good enough workaround, wouldn't it?

That workaround is great for me personally. These tests remain important
for production distributions.