Non-Vc builds broken (Re: [krita] /: Update krita to compile against Vc >= 1.0.0)
Fazekas László
mneko at freemail.hu
Sun Apr 10 14:23:11 UTC 2016
Hello,
I compiled Krita with Ubuntu 14.04 until the Vc change, then I wasn't
able to build it anymore. I completely dropped the 14.04 system and
installing 16.04 right now. Maybe it was my fault, but this change was a
big problem on my Linux too.
Fazek
2016-04-10 13:59 keltezéssel, Boudewijn Rempt írta:
> On Sun, 10 Apr 2016, Dmitry Kazakov wrote:
>
>> Erm... I would really expect that the version without Vc would go
>> into a separate branch, not doing any reverts in master. Now you made
>> all the people who already built Vc 1.2 rebuild Krita again and waste
>> about an hour of their time.
>
> Yes, I realize that. But:
>
> * CI was broken too
> * Windows builds from master were impossible
> * Raghukamath reported a big performance issue
>
> All together, that was enough for me to decide that we need to work on
> these issues in a branch and make master stable again.
>
> Boudewijn
>
>
>>
>> On Sat, Apr 9, 2016 at 1:39 PM, Boudewijn Rempt <boud at valdyas.org>
>> wrote:
>> Making the build work without Vc was too tricky for me, which
>> blocks building
>> Krita on Windows with MSVC. While I'm working to setup a
>> mingw-build system on
>> Windows, I've moved the Vc code to a new branch:
>>
>> rempt-port-vc
>>
>> Where we should work to fix building without Vc. In the
>> meantime, I've reverted
>> everything in master, so CI and Windows users can at least
>> build again.
>>
>> Also note this bug report:
>>
>> https://bugs.kde.org/show_bug.cgi?id=361546
>>
>> "The brushes have become really slow in recent 3.0 alpha builds
>> compared to their 2.9 counterparts"
>>
>> On Fri, 8 Apr 2016, Friedrich W. H. Kossebau wrote:
>>
>> Hi Thorsten,
>>
>> seems this breaks non-Vc builds of Krita (to be seen on
>> CI, which still has Vc
>> 0.7.4 and thus tries to build without), e.g.
>>
>> https://build.kde.org/job/krita%20master%20kf5-qt5/
>> PLATFORM=Linux,compiler=gcc/63/
>>
>> Making Vc a hard dep would long-term be perhaps the best,
>> but for now Vc 1.0
>> is not that spread yet, so would put extra burdens to new
>> contributors (&
>> distributors).
>>
>> See below for one thing I found already.
>>
>> Am Donnerstag, 7. April 2016, 12:56:24 CEST schrieb
>> Thorsten Zachmann:
>> Git commit b68c1c7666510e268babc4cd87cd9d5ad34ff2d5
>> by Thorsten Zachmann.
>> Committed on 07/04/2016 at 12:54.
>> Pushed by zachmann into branch 'master'.
>>
>> Update krita to compile against Vc >= 1.0.0
>>
>>
>> <snip>
>>
>> diff --git
>> a/libs/pigment/compositeops/KoVcMultiArchBuildSupport.h
>> b/libs/pigment/compositeops/KoVcMultiArchBuildSupport.h index
>> 09677a6..2b10257 100644
>> ---
>> a/libs/pigment/compositeops/KoVcMultiArchBuildSupport.h
>> +++
>> b/libs/pigment/compositeops/KoVcMultiArchBuildSupport.h
>> @@ -35,6 +35,7 @@
>> #pragma warning ( disable : 4244 )
>> #pragma warning ( disable : 4800 )
>> #endif
>> +#include <Vc/global.h>
>> #include <Vc/Vc>
>> #include <Vc/support.h>
>> #if defined _MSC_VER
>> @@ -44,11 +45,9 @@
>> #else /* HAVE_VC */
>>
>> namespace Vc {
>> - typedef enum {ScalarImpl} Implementation;
>> + typedef enum {ScalarImpl} CurrentImplementation;
>> }
>>
>>
>> There still is Vc::Implementation in the code, so should
>> this here be not a
>> change Implementation->CurrentImplementation, but rather
>> an addition of
>> CurrentImplementation?
>>
>> There is also Vc::CurrentImplementation::current() in the
>> general code, so the
>> typedef might not be the working solution.
>>
>> Cheers
>> Friedrich
>>
>>
>> --
>> Boudewijn Rempt | http://www.krita.org, http://www.valdyas.org
>> _______________________________________________
>> Krita mailing list
>> kimageshop at kde.org
>> https://mail.kde.org/mailman/listinfo/kimageshop
>>
>>
>>
>>
>> --
>> Dmitry Kazakov
>>
>>
>
>
>
> _______________________________________________
> Krita mailing list
> kimageshop at kde.org
> https://mail.kde.org/mailman/listinfo/kimageshop
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kimageshop/attachments/20160410/b8354444/attachment.html>
More information about the kimageshop
mailing list