3.5.6 'make check' problems (base)

[Matthew Woehlke]

David Faure wrote:
> On Monday 05 February 2007, Matthew Woehlke wrote:
>> David Faure wrote:
>>> On Monday 05 February 2007, Matthew Woehlke wrote:
>>>> command.cc:257: error: 'class SMTPProtocol' has no member named 
>>>> 'openPassDlg'
>>>>
>>>> ...but I can't figure out why. openPassDlg is a member of 
>>>> KIO::SlaveBase. SMTPProtocol is derived from KIO::TCPSlaveBase which in 
>>>> turn is derived from KIO::SlaveBase, so I don't see offhand what the 
>>>> problem is. Guesses?
>>> Because the test defines its own SMTPProtocol class, apparently.
>>> test_commands.cc:class SMTPProtocol {
>> Ack! I missed that, it's certainly the problem. Hmm, how to fix it... 
>> will have to think about that.
> 
> Ifdef'ing the call to openpassdlg when building that code as test, if I'm right
> that it's getting compiled twice?

That was one thought, although it seems pretty ugly to have '#if TEST' 
in release code. I would rather write a dummy function like e.g. the 
tests's SMTPProtocol::messageBox, but I haven't decided what needs to go 
in there.

Like I said, I'll look at it eventually. Right now I just broke 
KATE-part (oops!), so I'm otherwise occupied. :-)

Thanks for the pointers, though!

>> It would be very good I think to have 'make check' run along with 
>> regular builds, especially for KDE4. After all, WTF is the point of a 
>> test suite if the tests themselves are broken? ;-)
> 
> You're preaching to the choir here. 
> How many times do I need to say it? The whole unit test stuff has been
> completely overhauled for KDE4, and the tests are already compiled daily
> with the rest of the code.

Um, until I catch the "already" part. My apologies, I was remembering 
hearing 'we /will/ do this', not 'we /now are/ doing this'. :-)

-- 
Matthew
This message is non-smoking