18 Aug
2012
18 Aug
'12
2:16 p.m.
François, all,
I noticed that François has in his patches for Fedora incorporated commit
513ffc6e: [tdelibs] Add ability to set KLed off color. Into v3.5.13-sru
branch I am not putting this patch because it breaks the ABI.
I'm interested in your opinion: Add 513ffc6e to v3.5.13-sru or not?
Slavek
--
18 Aug
18 Aug
2:21 p.m.
New subject: [trinity-devel] Include 513ffc6e into v3.5.13-sru?
Le 18/08/2012 16:16, Slávek Banko a écrit :
François, all,
I noticed that François has in his patches for Fedora incorporated commit
513ffc6e: [tdelibs] Add ability to set KLed off color. Into v3.5.13-sru
branch I am not putting this patch because it breaks the ABI.
I'm interested in your opinion: Add 513ffc6e to v3.5.13-sru or not?
Slavek
Hello,
in 3.5.13-sru, kled is already crashes 3.5.13 non-rebuilt apps because
of a "tqled" => "led" renaming or something like that.
Ex: kpowersave 3.5.13 with tdelibs 3.5.13.1 crashes because of kled.
So I think it won't break it more to introduce this patch too.
Francois
3:30 p.m.
New subject: [trinity-devel] Include 513ffc6e into v3.5.13-sru?
On Saturday 18 of August 2012 16:21:48 Francois Andriot wrote:
Le 18/08/2012 16:16, Slávek Banko a écrit :
Let me disagree.
Can I mention again that, included "renaming" patches are:
+ Rename old tq methods that no longer need a unique name
+ Remove additional unneeded tq method conversions
+ Rename obsolete tq methods to standard names
+ Rename a few stragglers
+ Fix inadvertent "TQ" changes
These patches change such names, which anyway are in tqtinterface handled
using "#define" - for example: #define tqsetSizePolicy setSizePolicy. These
patches can not break the ABI. And not require recompilation of others.
Exceptions that require recompiling a few other packages I mentioned earlier:
http://trinity-devel.pearsoncomputing.net/?0::9413
513ffc6e patch is different from the above. This patch changes the ABI and
affects a large number of packages, which would have to be recompiled.
Therefore, this patch (though is very small) for me seems with a too big
impact than they should have a patches for SRU.
Slavek
--
François, all,
I noticed that François has in his patches for Fedora incorporated commit
513ffc6e: [tdelibs] Add ability to set KLed off color. Into v3.5.13-sru
branch I am not putting this patch because it breaks the ABI.
I'm interested in your opinion: Add 513ffc6e to v3.5.13-sru or not?
Slavek
Hello,
in 3.5.13-sru, kled is already crashes 3.5.13 non-rebuilt apps because
of a "tqled" => "led" renaming or something like that.
Ex: kpowersave 3.5.13 with tdelibs 3.5.13.1 crashes because of kled.
So I think it won't break it more to introduce this patch too.
Francois
6:14 p.m.
New subject: [trinity-devel] Include 513ffc6e into v3.5.13-sru?
Le 18/08/2012 17:30, Slávek Banko a écrit :
On Saturday 18 of August 2012 16:21:48 Francois
Andriot wrote:
Mmm I'd like to be as confident as you about the few number of
"exception" packages that should be rebuilt.
But anyway, you are right about this particular patch, adding it to
3.5.13 causes too much troubles.
So I've simply removed it from tdelibs.
I understand that, theorically, the "TQ" renaming keeps full compatibility.
But I've seen that this renaming has more side effects (positive or
negative) than we could think.
For example, the original 3.5.13 digikam does not work at all (at least
on my computer). See bug 963.
After applying the "renaming" patches from SRU, it magically works.
In that case, it is obviously a good news. But this was still unexpected.
So, to avoid any bad surprise, I still intend to rebuild every packages
for 3.5.13.1 .
Which I hope will be released soon :-) (this is a subliminal message)
Francois
Le 18/08/2012 16:16, Slávek Banko a écrit :
Let me disagree.
Can I mention again that, included "renaming" patches are:
+ Rename old tq methods that no longer need a unique name
+ Remove additional unneeded tq method conversions
+ Rename obsolete tq methods to standard names
+ Rename a few stragglers
+ Fix inadvertent "TQ" changes
These patches change such names, which anyway are in tqtinterface handled
using "#define" - for example: #define tqsetSizePolicy setSizePolicy. These
patches can not break the ABI. And not require recompilation of others.
Exceptions that require recompiling a few other packages I mentioned earlier:
http://trinity-devel.pearsoncomputing.net/?0::9413
513ffc6e patch is different from the above. This patch changes the ABI and
affects a large number of packages, which would have to be recompiled.
Therefore, this patch (though is very small) for me seems with a too big
impact than they should have a patches for SRU.
Slavek François, all,
I noticed that François has in his patches for Fedora incorporated commit
513ffc6e: [tdelibs] Add ability to set KLed off color. Into v3.5.13-sru
branch I am not putting this patch because it breaks the ABI.
I'm interested in your opinion: Add 513ffc6e to v3.5.13-sru or not?
Slavek
Hello,
in 3.5.13-sru, kled is already crashes 3.5.13 non-rebuilt apps because
of a "tqled" => "led" renaming or something like that.
Ex: kpowersave 3.5.13 with tdelibs 3.5.13.1 crashes because of kled.
So I think it won't break it more to introduce this patch too.
Francois
19 Aug
19 Aug
5:49 a.m.
New subject: [trinity-devel] Include 513ffc6e into v3.5.13-sru?
On 08/18/2012 10:30 AM, Slávek Banko wrote:
Let me disagree.
Can I mention again that, included "renaming" patches are:
+ Rename old tq methods that no longer need a unique name
+ Remove additional unneeded tq method conversions
+ Rename obsolete tq methods to standard names
+ Rename a few stragglers
+ Fix inadvertent "TQ" changes
These patches change such names, which anyway are in tqtinterface handled
using "#define" - for example: #define tqsetSizePolicy setSizePolicy. These
patches can not break the ABI. And not require recompilation of others.
Exceptions that require recompiling a few other packages I mentioned earlier:
http://trinity-devel.pearsoncomputing.net/?0::9413
513ffc6e patch is different from the above. This patch changes the ABI and
affects a large number of packages, which would have to be recompiled.
Therefore, this patch (though is very small) for me seems with a too big
impact than they should have a patches for SRU.
Slavek
Francios, Slavek,
I do not understand the issue at the ABI level, but I would advise only
implementing the *necessary* name changes. I have had fits with the k->t renames
on numerous files breaking builds that are frustrating to fix. I don't mind
doing it if the change is necessary, but wholesale renaming just for the sake of
it should be avoided.
I will defer to your wisdom on this, but we absolutely need kpowersave to
work. I don't mind rebuilding, so if it makes sense do it. But the foundation of
3.5.13-1 should be as close to 3.5.13 as possible...
--
David C. Rankin, J.D.,P.E.
4277
days inactive
4278
days old
4 comments
3 participants
participants (3)
-
David C. Rankin -
Francois Andriot -
Slávek Banko