Re: [lmi] wx_test_paste_census.cpp

[Vadim Zeitlin]

On Fri, 12 Dec 2014 23:36:29 +0000 Greg Chicares <address@hidden> wrote:

GC> // ERASE THIS BLOCK COMMENT WHEN IMPLEMENTATION COMPLETE. The block
GC> // comment below changes the original specification, and does not
GC> // yet describe the present code. Desired changes:
GC> //  - Save pastable data inline; don't extract from user manual.
GC> //  - Validate all columns after each step (after initial pasting).
GC> //  - Test change in class defaults (in addition to case defaults).
GC> //  - Place the saved file in 'gui_test_path'.

 Hello,

 The (first) attached patch implements the changes above. It also improves
error reporting (i.e. more information is now given in case of test
failure) and corrects the URL to the paste example in the manual.

 Unfortunately it also requires you to update the version of wxWidgets you
use once again, this time to c62cbe89708cc26d0f86bd424122e3969387494d which
fixes the problem with not being able to emulate Ctrl-Alt-E key combination
(which is used for editing the class defaults) using wxUIActionSimulator,
as you can see at

        
https://github.com/wxWidgets/wxWidgets/commit/c62cbe89708cc26d0f86bd424122e3969387494d

FWIW the previous commit is also lmi-related and fixes a crash which would
happen if you, as the user, chose "Cancel" in the message box asking
whether the document should be saved if something went wrong during the
test. I will soon submit another patch which should suppress this message
box in the first place, but it never hurts to have one potential crash less.


 The other attached patches try to provide solutions for both possible
answers from Kim concerning whether PasteCensus.cns file should be
preserved or not. Please notice that they use output_file_existence_checker
and so depend on the patches in PDF creation test series from the message at

        http://lists.nongnu.org/archive/html/lmi/2014-12/msg00089.html

The second patch assumes that the file does not need to be preserved and
ensures that it is destroyed. The third one does preserve the file.

 You may notice that the net effect of applying both the second and third
patches is the same as not applying neither of them. However I'd prefer to
consistently use output_file_existence_checker everywhere, even if it's not
very useful with the third patch. In case you disagree, please skip
applying these patches, of course.

 And if I still didn't manage to confuse you entirely, here is my last
attempt to do it in the form of the decision table showing which patches
should be applied depending on different answers:


        PasteCensus.cns should be|
        ------------------------ |             |           |
        Greg agrees with me     \|  Preserved  |  Deleted  |
        _________________________|_____________|___________|
                                 |             |           |
                Yes              |     2+3     |     2     |
                No               |     none    |     2     |
       __________________________|_____________|___________|

 Please let me know if you have any questions and thanks in advance,
VZ

0001-Update-census-paste-test-to-correspond-to-the-revise.patch
Description: Text document

0002-Use-output_file_existence_checker-helper-in-census-p.patch
Description: Text document

0003-Allow-preserving-files-used-with-output_file_existen.patch
Description: Text document