[KDE/Mac] OSX/CI: Uses of outdated Q_WS_MAC in KF5 projects

Ian Wadham iandw.au at gmail.com
Sat Jan 24 00:41:58 UTC 2015

On 24/01/2015, at 9:13 AM, David Faure wrote:
> On Saturday 24 January 2015 09:10:33 Ian Wadham wrote:
>> Hello David and Marko,
>> On 24/01/2015, at 4:28 AM, Marko Käning wrote:
>>> On 23 Jan 2015, at 18:18 , David Faure <faure at kde.org> wrote:
>>>> IMHO submit specific patches or file specific tickets, not a catch-all
>>>> ticket like "there are still 191 instances of Q_WS_MAC in qtbase", which
>>>> nobody will ever feel like fixing in one go…
>> @Marko: About how many instances are there actually?  I hope it is not
>>                  anywhere near 191...
> I didn't make up that number, I used git grep Q_WS_MAC | wc -l

My warmest apologies for doubting that, David.

I wonder if the equivalent of "grep -Rl" would help.  If the occurrences are
clustered in modules to provide Qt 4 support during porting, no problem,
but if they are in core or GUI modules, someone at Qt should take a look.

Unfortunately, I cannot get the source of Qt 5.4.  I tried a few weeks back,
but my (rather slow) Aussie broadband died in the attempt.

I did grep -Rl * on the bits of Qt 5.4 I *do* have and found 68 affected files…
Curiously, there were 213 hits on "grep -R *", more than David got… :-(

Just as an aperitif, have a look at:
Search through those sources for Q_.._MAC.  There is a mixture of uses
of Q_OS_MAC and Q_WS_MAC, but *most* uses are Q_WS_MAC.

Many Qt widgets were also affected, in .h and/or .cpp files, e.g.
QPushButton, QLineEdit and QMainWindow.

I found the affected modules of Qt (over 60 files involved) were:
but there could be more.  I have an incomplete set of Qt 5.4 modules.

>> @David: "191" and "nobody will ever feel like" should not enter into it. 
>> Digia is a company and has commercial Qt customers and paid staff, IIUC.
> They do, but if you report it then it's not a request from a commercial Qt 
> customer.
>>                 We should not be depending on volunteers from KDE and
>> KDE-Mac or Qt users' groups for this.  Surely it is in Digia's and Qt's
>> commercial interests to investigate, as a matter of priority, what impact
>> the Q_WS_MAC issue might be having on the Qt-Mac 5 product?  If it is not
>> serious, well and good: people can fix the instances as they go along.  But
>> if the impact on Qt-Mac 5 globally *is* serious, somebody should get onto
>> it right away.  Marko or René can report a bug or two "through channels",
>> but I doubt if that can achieve urgent action if it turns out that the
>> problem is serious and affects the whole Qt-Mac 5 product.
> That's my point - report the actual impact (bugs and missing features),
> not the grep result.

See above for starters.  I have no idea what the impact is, because I am not
an expert on the Qt 5.4 source code.  I can see that quite sizeable chunks
would be omitted from the QWidget code in Qt-Mac 5.4 and I am surprised
that Marko can run any KF5 apps at all on Apple OS X.  It needs an expert
to look at the code, I would say, but it certainly *looks* wrong.

David, I used to be Technical Director of a software company about the
size of KDAB.  Then I became a freelance consultant to government and
industry.  Wherever I went, I would try to cultivate relationships with key
players in our supplier companies and it payed off handsomely on those
occasions when something big went wrong.  Suppliers and customers
both benefited.

I was just wondering if you or Laurent could have a quiet word to someone
you know at Digia, so that an expert could be assigned to spend an hour or
so looking at this problem.  It would save all of us a lot of time and, if the problem
is serious and widespread, then it could save Digia some grief when the use of
Qt-Mac 5 builds up.  Just an idea… :-)

All the best, Ian W.

More information about the kde-mac mailing list