A few days ago I was investigating a bug in the EFL WebKit port and
noticed WebKit's and Evas' handling of Fontconfig are somewhat
incompatible: while the evas_font code calls both FcInit() and FcFini()
when on initialization and shutdown, respectively, WebKit keeps some
Fontconfig objects alive un
On 30/01/13 17:15, Raphael Kubo da Costa wrote:
> A few days ago I was investigating a bug in the EFL WebKit port and
> noticed WebKit's and Evas' handling of Fontconfig are somewhat
> incompatible: while the evas_font code calls both FcInit() and FcFini()
> when on initialization and shutdown, res
Tom Hacohen writes:
> You only removed code, are you sure we still init everything after your
> changes?
>
> (Sorry, I don't have time to take a look yet).
I've tried both expedite and loading a few web pages with different
fonts and everything works fine -- the other libraries I mentioned do
no
On 30/01/13 17:29, Raphael Kubo da Costa wrote:
> Tom Hacohen writes:
>
>> You only removed code, are you sure we still init everything after your
>> changes?
>>
>> (Sorry, I don't have time to take a look yet).
>
> I've tried both expedite and loading a few web pages with different
> fonts and ev
Tom Hacohen writes:
> ==24639== 736 (512 direct, 224 indirect) bytes in 2 blocks are
> definitely lost in loss record 152 of 189
[...]
> Evas leaks now...
I get different results here with the attached program. Namely, the
blocks are marked as still reachable instead of definitely lost. This is
On Wed, 30 Jan 2013 20:06:19 +0200 Raphael Kubo da Costa
wrote:
> Tom Hacohen writes:
>
> > ==24639== 736 (512 direct, 224 indirect) bytes in 2 blocks are
> > definitely lost in loss record 152 of 189
> [...]
> > Evas leaks now...
>
> I get different results here with the attached program. Nam
Just use "elementary_test", no need to get fancy and write your own
applications... :)
As for solving the issue at hand: I don't quite understand, does webkit
init and shutdown evas all the time? Why doesn't webkit init and shutdown
fontconfig as well?
If they are using it, they should init and sh
Tom Hacohen writes:
> Just use "elementary_test", no need to get fancy and write your own
> applications... :)
Same results: no {definitely,indirectly,possibly} lost objects, some
kilobytes of still reachable objects.
> As for solving the issue at hand: I don't quite understand, does webkit
> i
On Wed, 30 Jan 2013 19:15:20 +0200 Raphael Kubo da Costa
said:
sounds to me like a tonne of other libs are not being "nice" with fc. :) ie
knowing to call fcini/fini.
evas does this so valgrind stops bitching about not freeds/still reachable
stuff on application shutdown - it reduces the noise.
Carsten Haitzler (The Rasterman) writes:
> so as a result, evas is following a much stricter clean "init and shtudown
> when
> u are done" model which at least allows you to write an app, use evas, then
> stop using it and for the next 10 weeks run it without ever using evas again
> and
> have
On Thu, 31 Jan 2013 02:41:28 +0200 Raphael Kubo da Costa
said:
> Carsten Haitzler (The Rasterman) writes:
>
> > so as a result, evas is following a much stricter clean "init and shtudown
> > when u are done" model which at least allows you to write an app, use evas,
> > then stop using it and f
On Wednesday, January 30, 2013, Raphael Kubo da Costa wrote:
> Carsten Haitzler (The Rasterman) >
> writes:
>
> > so as a result, evas is following a much stricter clean "init and
> shtudown when
> > u are done" model which at least allows you to write an app, use evas,
> then
> > stop using it an
On 31/01/13 01:20, Gustavo Sverzut Barbieri wrote:
> On Wednesday, January 30, 2013, Raphael Kubo da Costa wrote:
>
>> Carsten Haitzler (The Rasterman) >
>> writes:
>>
>>> so as a result, evas is following a much stricter clean "init and
>> shtudown when
>>> u are done" model which at least allows
On Thursday, January 31, 2013, Tom Hacohen wrote:
> On 31/01/13 01:20, Gustavo Sverzut Barbieri wrote:
>
>> On Wednesday, January 30, 2013, Raphael Kubo da Costa wrote:
>>
>> Carsten Haitzler (The Rasterman) >
>>> writes:
>>>
>>> so as a result, evas is following a much stricter clean "init and
Gustavo Sverzut Barbieri writes:
> Just dont call evas_shutdown()?
Not calling evas_shutdown() is also an option, however that would mean
not shutting another library down properly, with all the risks and
problems that brings.
> Or call it after you know every other library using it is
> done/f
Gustavo Sverzut Barbieri writes:
> On Thursday, January 31, 2013, Tom Hacohen wrote:
>> Yes, that's what I said, so obviously I agree with that. Fix webkit.
Talk is cheap :-)
> Kubo: do you know if FcIni/FcFini are refcounted like EFL?
They are not. If the general perception here is that Evas
On Thu, Jan 31, 2013 at 9:18 AM, Raphael Kubo da Costa
wrote:
> Gustavo Sverzut Barbieri writes:
>> Kubo: do you know if FcIni/FcFini are refcounted like EFL?
>
> They are not. If the general perception here is that Evas should not be
> changed, trying to talk to the Fontconfig guys and change
>
Gustavo Sverzut Barbieri writes:
> ouch, so they are not refcounted and it will free at first exit...
> that's bad, quite bad :-)
>
> Option #1 is talk to fc guys and see if they can add that. It would be
> more correct.
Looking at Fontconfig's mailing list archives, it looks like vtorri hit
the
On Thu, 31 Jan 2013 15:40:50 +0200 Raphael Kubo da Costa
said:
> Gustavo Sverzut Barbieri writes:
>
> > ouch, so they are not refcounted and it will free at first exit...
> > that's bad, quite bad :-)
> >
> > Option #1 is talk to fc guys and see if they can add that. It would be
> > more correc
But then valgrind complains, i.e there's noise. :( That's also bad for the
whole of us.
On Fri, Feb 1, 2013 at 2:25 PM, Carsten Haitzler wrote:
> On Thu, 31 Jan 2013 15:40:50 +0200 Raphael Kubo da Costa
> said:
>
> > Gustavo Sverzut Barbieri writes:
> >
> > > ouch, so they are not refcounted a
On Fri, Feb 1, 2013 at 12:21 PM, Tom Hacohen wrote:
> But then valgrind complains, i.e there's noise. :( That's also bad for the
> whole of us.
>
if (getenv("VALGRINDING_LOLOLOLOLOL")) FcFini();
>
> On Fri, Feb 1, 2013 at 2:25 PM, Carsten Haitzler wrote:
>
>> On Thu, 31 Jan 2013 15:40:50 +0200 R
Tom Hacohen writes:
> But then valgrind complains, i.e there's noise. :( That's also bad for the
> whole of us.
Valgrind complaining about reachable blocks shouldn't as bad as actual
leaks anyway.
--
Everyone hates slo
valgrind is complaining about "definitely lost", otherwise I wouldn't have
cared that much.
On Fri, Feb 1, 2013 at 2:48 PM, Raphael Kubo da Costa <
raphael.kubo.da.co...@intel.com> wrote:
> Tom Hacohen writes:
>
> > But then valgrind complains, i.e there's noise. :( That's also bad for
> the
>
Carsten Haitzler (The Rasterman) writes:
> that's why efl does it by default... to cover "this case". we can stop calling
> it.. but there are downsides to that.. :) i'm on the fence - it's better to
> call it, BUT... if its not practical.. then commenting it out is ok i guess.
After doing some
I'll talk to Behdad about it once he goes on-line and I'll ask him about it.
This whole things smells. :)
On Fri, Feb 1, 2013 at 2:53 PM, Raphael Kubo da Costa <
raphael.kubo.da.co...@intel.com> wrote:
> Carsten Haitzler (The Rasterman) writes:
>
> > that's why efl does it by default... to cove
Tom Hacohen writes:
> valgrind is complaining about "definitely lost", otherwise I wouldn't have
> cared that much.
With efl/ and elementary/ at r83566 and valgrind 3.8.1, running
elementary_test with my patch does not show any definitely lost blocks
(which makes sense, since all those Fontconfi
Well. I just ran elementary test with your patch and without and it and
shows leaks.
If others can confirm it's not the case, and it's just my setup that is
broken, I'm fine with your patch..
On Fri, Feb 1, 2013 at 3:04 PM, Raphael Kubo da Costa <
raphael.kubo.da.co...@intel.com> wrote:
> Tom Ha
Ok, just had a short chat with Behdad.
He says that you are right and fontconfig should not be init and shutdown.
I disagree with this design choice, but apparently
that's how it is in fontconfig so arguing against it or disagreeing will
get us nowhere. ;)
Either someone will apply this patch, or I
On Sat, Feb 2, 2013 at 7:28 AM, Tom Hacohen wrote:
> Ok, just had a short chat with Behdad.
> He says that you are right and fontconfig should not be init and shutdown.
> I disagree with this design choice, but apparently
> that's how it is in fontconfig so arguing against it or disagreeing will
>
On Sat, 2 Feb 2013 14:36:12 +0900 Cedric BAIL said:
> On Sat, Feb 2, 2013 at 7:28 AM, Tom Hacohen wrote:
> > Ok, just had a short chat with Behdad.
> > He says that you are right and fontconfig should not be init and shutdown.
> > I disagree with this design choice, but apparently
> > that's how
On Sat, 2 Feb 2013 15:01:39 +0900 Carsten Haitzler (The Rasterman)
wrote:
> On Sat, 2 Feb 2013 14:36:12 +0900 Cedric BAIL
> said:
>
> > On Sat, Feb 2, 2013 at 7:28 AM, Tom Hacohen wrote:
> > > Ok, just had a short chat with Behdad.
> > > He says that you are right and fontconfig should not be
Cedric BAIL writes:
> On Sat, Feb 2, 2013 at 7:28 AM, Tom Hacohen wrote:
>> Ok, just had a short chat with Behdad.
>> He says that you are right and fontconfig should not be init and shutdown.
>> I disagree with this design choice, but apparently
>> that's how it is in fontconfig so arguing agai
Cedric Bail
On Feb 3, 2013 2:07 AM, "Raphael Kubo da Costa" <
raphael.kubo.da.co...@intel.com> wrote:
>
> Cedric BAIL writes:
>
> > On Sat, Feb 2, 2013 at 7:28 AM, Tom Hacohen wrote:
> >> Ok, just had a short chat with Behdad.
> >> He says that you are right and fontconfig should not be init and
On Fri, Feb 1, 2013 at 8:28 PM, Tom Hacohen wrote:
> Ok, just had a short chat with Behdad.
> He says that you are right and fontconfig should not be init and shutdown.
> I disagree with this design choice, but apparently
> that's how it is in fontconfig so arguing against it or disagreeing will
>
Could you please backport it as well?
On Mon, Feb 4, 2013 at 3:39 PM, Bruno Dilly wrote:
> On Fri, Feb 1, 2013 at 8:28 PM, Tom Hacohen wrote:
> > Ok, just had a short chat with Behdad.
> > He says that you are right and fontconfig should not be init and
> shutdown.
> > I disagree with this des
On Mon, Feb 4, 2013 at 1:51 PM, Tom Hacohen wrote:
> Could you please backport it as well?
>
Sure, it's done.
Thanks
>
> On Mon, Feb 4, 2013 at 3:39 PM, Bruno Dilly wrote:
>
>> On Fri, Feb 1, 2013 at 8:28 PM, Tom Hacohen wrote:
>> > Ok, just had a short chat with Behdad.
>> > He says that you
Cool, thanks!
On Mon, Feb 4, 2013 at 4:07 PM, Bruno Dilly wrote:
> On Mon, Feb 4, 2013 at 1:51 PM, Tom Hacohen wrote:
> > Could you please backport it as well?
> >
>
> Sure, it's done.
> Thanks
>
> >
> > On Mon, Feb 4, 2013 at 3:39 PM, Bruno Dilly
> wrote:
> >
> >> On Fri, Feb 1, 2013 at 8:28
37 matches
Mail list logo
