> while ((i<2000)); do
>  echo "awful.client.focus.byidx( 1)
>  if client.focus then client.focus:raise() end
> " | awesome-client; ((i++)); done

This isn't that kind of leak. I tried such tests with no success.

ava...@aliso ~ % ps aux | grep awesome
avatar   24869  0.7  0.3 194460 15324 ?        Ss   Sep15   0:06 awesome

ava...@aliso ~ % while ((i<2000)); do
ava...@aliso ~ while>  echo "awful.client.focus.byidx( 1)
ava...@aliso ~ while dquote>  if client.focus then client.focus:raise() end
ava...@aliso ~ while dquote> " | awesome-client; ((i++)); done

ava...@aliso ~ % ps aux | grep awesome
avatar   24869  1.0  0.3 194780 15664 ?        Ss   Sep15   0:10 awesome

I think that may be issue in  cairo/lua, that activates with complex
conditions. And I have no idea how to search them. Valgrind just
crashes with awesome in my configuration, dmalloc isn't very accurate

2010/9/16, Javier Barroso <[email protected]>:
> Hi,
> On Wed, Sep 15, 2010 at 6:40 PM, [AvataR] <[email protected]> wrote:
>> ... and again
>>
>> ava...@aliso ~ % uptime
>>  19:39:23 up 2 days, 11:44,  1 user,  load average: 1.18, 1.10, 0.84
>>
>> ava...@aliso ~ % uname -m
>> x86_64
>>
>> ava...@aliso ~ % ps axo rsz,command | egrep "awesome|tomcat"
>> 33240 /opt/sun-jdk-1.6.0.21/bin/java ...
>> 126952 awesome
>>
>> WTF
>>
>> Hey! May be awesome should be rewritten in java? It will be more
>> efficient ))
> If you do it, please name it "unamazing"
>
> So, lets create rc.java config file in java ... you will have
> nightmares with it 3-)
>
> Now seriously,
>
> Maybe you could create a test which we could replay in other distros
> (maybe a gentoo library issue ?). I tried something like:
>
> while ((i<2000)); do
>  echo "awful.client.focus.byidx( 1)
>   if client.focus then client.focus:raise() end
> " | awesome-client; ((i++)); done
>
> which simulate 2000 Mod4-j keystrokes
>
> And watching ps -Cawesome -o rss= in a on top windows, but no memory
> leak is observed here in debian sid amd64.
>
> Regards,
>
> --
> To unsubscribe, send mail to [email protected].
>

--
To unsubscribe, send mail to [email protected].

Reply via email to