Yeah, in fact my workaround was not hovering the timeline so my pc is happy. 
Hovering for a moment, tipically to move the pointer to the shelf, doesn't 
trigger the issue, it starts only when sitting on the timeline for some tenth 
of a second.

Massimo


Il 22 agosto 2016 02:00:41 CEST, Al Poole <[email protected]> ha scritto:
>When you hover over the timeline to see a thumbnail that triggers the
>thumbnail creation.
>
>Keep going back over up and down the timeline and that is triggerign
>loads
>of new rage_thumb processes more than the limit set.
>
>So as a user I'm like "Where's my thumbnails, they are black...so let
>me go
>up and down looking for them, but they are still not there and at the
>same
>time WTF my computer isn't happy at all..."
>
>on mouse over the timeline/scroller lots of new rage_thumb being
>created...
>
>I think that is what is happening
>
>On Mon, Aug 22, 2016 at 12:28 AM, Carsten Haitzler
><[email protected]>
>wrote:
>
>> On Sun, 21 Aug 2016 19:47:20 +0200 Massimo Maiurana
><[email protected]>
>> said:
>>
>> > Carsten Haitzler (The Rasterman) ha scritto il 21/08/2016 alle
>15:05:
>> > > On Sun, 21 Aug 2016 14:22:33 +0200 Massimo Maiurana <
>> [email protected]>
>> > > said:
>> > >
>> > >> Carsten Haitzler (The Rasterman) ha scritto il 21/08/2016 alle
>14:03:
>> > >>> On Sun, 21 Aug 2016 10:42:47 +0200 Massimo Maiurana <
>> [email protected]>
>> > >>> said:
>> > >>>
>> > >>>> Carsten Haitzler (The Rasterman) ha scritto il 21/08/2016 alle
>> 01:42:
>> > >>>>> On Sat, 20 Aug 2016 19:04:44 +0200 Massimo Maiurana <
>> [email protected]>
>> > >>>>> said:
>> > >>>>>
>> > >>>>>> Carsten Haitzler (The Rasterman) ha scritto il 20/08/2016
>alle
>> 14:17:
>> > >>>>>>> On Sat, 20 Aug 2016 09:41:00 +0200 Massimo Maiurana
>> > >>>>>>> <[email protected]> said:
>> > >>>>>>>
>> > >>>>>>>> There is an annoying issue that I found in rage from some
>> months ago
>> > >>>>>>>> but I've always forgot to report it here, sorry :(
>> > >>>>>>>>
>> > >>>>>>>> There are cases where rage_thumb just go crazy, spawns
>several
>> > >>>>>>>> processes, eats up the cpu and reamins active even if I
>close
>> rage,
>> > >>>>>>>> so I have to "killal -9 rage_thumb". Tipically it happens
>if I
>> > >>>>>>>> navigate the movie clicking with the mouse in the progress
>bar,
>> so
>> > >>>>>>>> in order to avoid it I use to navigate via arrow keys.
>> > >>>>>>>
>> > >>>>>>> well - it's generating thumbanils for the whole movie
>(every 10
>> > >>>>>>> seconds or so it snapshots). it literally sits and loads
>the
>> movie
>> > >>>>>>> relying on the gst generic evas loader to load a specific
>frame #
>> > >>>>>>> from the video file. it loads, scales down, and saves to a
>big
>> eet
>> > >>>>>>> archive with all the thumbs in it for the video...
>> > >>>>>>>
>> > >>>>>>> why is it spinning? some bakctrace, some trace/debug as to
>what
>> it's
>> > >>>>>>> spinning on etc.... would be useful. :)
>> > >>>>>>
>> > >>>>>> I would like to help but need also help on how to generate
>> > >>>>>> backtraces :)
>> > >>>>>
>> > >>>>> gdb /usr/local/lib/rage/utils/rage_thumb `pidof rage_thumb`
>> > >>>>>
>> > >>>>> ...
>> > >>>>>
>> > >>>>> bt
>> > >>>>>
>> > >>>>> :)
>> > >>>>>
>> > >>>>> also getting the file it is choking on would be nice. the
>file
>> will be
>> > >>>>> on the cmdline passed to rage_thumb..
>> > >>>>
>> > >>>> Ok, I did this: I opened a movie, clicked on some points in
>the
>> progress
>> > >>>> bar to go forward and the cpu went 100% (both cores). At this
>point
>> I
>> > >>>> had six instances of rage_thumb running. This is the command
>line:
>> > >>>> rage
>/home/max/Video/Il.Volo.Del.Falco.2014.iTALiAN.BDRiP.XviD-
>> HDi.avi
>> > >>>>
>> > >>>> Then I closed rage but all rage_thumb processes still was
>alive and
>> > >>>> kicking... too kicking :)
>> > >>>> Htop reported small cpu usage for rage_thumb, every process
>was
>> eating
>> > >>>> just around 1/1.5% of it. Top made me see that there were many
>> > >>>> evas_image_load processes that instead was eating much more
>cpu,
>> 10% and
>> > >>>> occasionally 20% each process. I generated a backtrace from
>one of
>> the
>> > >>>> rage_thumb processes, I don't know how it is useful but it is
>> attached.
>> > >>>> In the meantime all processes quitted, so it is just a matter
>of
>> time
>> > >>>> waiting they do their job, but the problem is that this jog is
>cpu
>> and
>> > >>>> time consuming :)
>> > >>>>
>> > >>>> Should I run gdb on evas_image_load and generate a backtrace?
>> > >>>
>> > >>> doesn't sound hung to me. it sounds like its actually working.
>it
>> may take
>> > >>> several minutes to thumbnail the video depending on how long it
>is...
>> > >>
>> > >> It doesn't hang but it makes my pc unresponsive for a long time,
>even
>> > >
>> > > no - i meant that rage_thumb isn't hung - it sounds by your
>> description,
>> > > like it's doing it's job. it's trying to generate thumbnails...
>> > >
>> > > i wonder if the issue is that the stream cannot be seeked and
>thus it's
>> > > simply repeatedly generating frame 0 because a seek to a point
>fails...
>> > >
>> > > if you do:
>> > >
>> > > strace -f -o ~/log -p `pidof rage_thumb`
>> > >
>> > > then grep through the file for "exec" - you should find the exec
>lines
>> that
>> > > run the generic loader... what are the aerguments? there should
>be the
>> file
>> > > plus a number. the number should be large - eg going up by 1000
>per
>> second
>> > > being snapped. it will try and get an image for every 10 seconds
>from
>> > > memory so number should be going up by 10k each run). maybe check
>the
>> > > strace log and see what the child processes are exiting with (0
>for
>> > > success? something else?).
>> >
>> > On the terminal where I ran strace I saw that it keeps spawning new
>> > processes, see a part of the output:
>> > strace: Process 31513 attached
>> > strace: Process 31514 attached
>> > strace: Process 31515 attached
>> > strace: Process 31516 attached
>> > strace: Process 31517 attached
>> >
>> > Consider that this is the end of the output, the first pid was
>22042.
>> >
>> > The command should be this:
>> > execve("/opt/e17/lib/evas/utils/evas_image_loader.avi",
>> > ["/opt/e17/lib/evas/utils/evas_ima"...,
>> > "/home/max/Video/Boulevard.2014.B"..., "-head", "-key", "3340000"],
>[/*
>> > 65 vars */] <unfinished ...>
>>
>> well thats at about 55 minutes... so if that video is about 55 or so
>> minutes
>> long... then it actually finished - right? processes went away. they
>> finished
>> thumbnailing. they just took a while... you should have seen a LOt of
>> execve
>> calls with different params...
>>
>> so the issue is more that you have 6 processes around. there is a
>limit.
>> why
>> isn't it working?
>>
>> > >> after I close rage. FWIU there should be a limit in how much cpu
>the
>> > >> thumbnailing processes can take, or a limit in the number of
>> > >
>> > > it sets the priority of the video thumb to 10 - the higher the
>priority
>> > > value the less likely the system will be to assign cpu to it - or
>> that's
>> > > how it should work. this should be inherited by child processes
>too.
>> this
>> > > is the nice value fyi (it uses setpriority() - see manual page):
>> > >
>> > > "The default nice value is {NZERO}; lower nice values shall 
>cause
>> more
>> > > favorable   scheduling."
>> > >
>> > >> thumbnailing processes which can be run at the same time, or at
>the
>> very
>> > >> least an option to turn off thumbnailing. I understand that my
>PC is
>> > >> quite old (6 years), but I would like to keep using it as much
>as
>> > >> possible :)
>> > >
>> > > i'm not going to make it an option. thumbnailing of videos is
>central
>> ro
>> > > rage and being able to know what on earth you are dealing with.
>you
>> haven't
>> > > seen the browser mode yet then where it'll be plainly obvious
>that no
>> > > thumbnail makes the ui pretty poor. it's central tot he ui
>design.
>> >
>> > Indeed I'm using rage from quite a long time and thumbnailing has
>always
>> > been there, this problem started recently so I guess there is no
>point
>> > in disabling thumbnailing :)
>>
>> yeah. i'd like to fix and know what's going on rather than stick my
>head
>> int he
>> sand and "make it an option" to turn it off.
>>
>> > >>>>>> If it would segvs i would do it via a "bt" command in gdb,
>but it
>> > >>>>>> doesn't crash and also rage_thumb keeps running in multiple
>> processes
>> > >>>>>> when I stop rage, so how should I trace its activities?
>> > >>>>>
>> > >>>>> ^^^^^ you can attach to an existing process any time. when
>you
>> attach
>> > >>>>> the process will be paused. you can grab a bt then. you can
>let it
>> > >>>>> continue running with "c" (continue) cmd any time. you can
>pause it
>> > >>>>> again with ctrl +c in gdb an get another backtrace. switch
>stack
>> frame
>> > >>>>> with "fr N" where N matches the backtrace number. list the
>code
>> where
>> > >>>>> you are at with "l". list a specific line number in that file
>in
>> > >>>>> context with "l 1234". print a variable you see with "p
>varname"
>> > >>>>> dereference a pointer (like c/c++ does all the time) with "p
>> *varname"
>> > >>>>> or if its a struct geta  member with "p varname->member". gdb
>will
>> even
>> > >>>>> be able to tab complete these like a shell does with files.
>:)
>> > >>>>>
>> > >>>>> learning gdb is a very powerful thing. you can find out so
>much
>> useful
>> > >>>>> info about a process... :)
>> > >>>>>
>> > >>>>>>>
>> > >>>>>>>> Thanks
>> > >>>>>>>> Massimo
>> > >>>>>>>>
>> > >>>>>>>> Carsten Haitzler (The Rasterman) ha scritto il 20/08/2016
>alle
>> 05:45:
>> > >>>>>>>>> So here is a new release of Rage.
>> > >>>>>>>>>
>> > >>>>>>>>> It is a simple video and audio player intended to be
>slick yet
>> > >>>>>>>>> simplistic, much like Mplayer. You can provide 1 or more
>files
>> to
>> > >>>>>>>>> play on the command-line or just DND files onto the rage
>> window to
>> > >>>>>>>>> insert them into the playlist. You can get a visual
>> representation
>> > >>>>>>>>> of everything on the playlist by hitting the / key, or
>just
>> > >>>>>>>>> hovering your mouse over the right side of the window.
>Mouse
>> back
>> > >>>>>>>>> over the left side of the window to dismiss it or press
>the key
>> > >>>>>>>>> again. It has a full complement of key controls if you
>see the
>> > >>>>>>>>> README for the full list. It will automatically search
>for
>> album
>> > >>>>>>>>> art for music files, if not already cached, and display
>that.
>> It
>> > >>>>>>>>> even generates thumbnails for the timeline of a video and
>> allows
>> > >>>>>>>>> you to preview the position on mouseover of the position
>bar
>> at the
>> > >>>>>>>>> bottom of the window.
>> > >>>>>>>>>
>> > >>>>>>>>> If you run it without any files as arguments, it will go
>into
>> > >>>>>>>>> "browser" mode where it will index $HOME/Videos and
>basically
>> become
>> > >>>>>>>>> a simple media center.
>> > >>>>>>>>>
>> > >>>>>>>>> A feature list at this point:
>> > >>>>>>>>>
>> > >>>>>>>>>   * Play video and audio files
>> > >>>>>>>>>   * Support a playlist via command-line
>> > >>>>>>>>>   * Insert to playlist via DND
>> > >>>>>>>>>   * Controls hide on mouse idle, and appear on mouse
>movement
>> > >>>>>>>>>   * Fullscreen mode support with automatic "no blank"
>support
>> > >>>>>>>>>   * Playlist visual previews and controls
>> > >>>>>>>>>   * Subtitle file support
>> > >>>>>>>>>   * Supports Gstreamer 0.10, Gstreamer 1.x, Xine and VLC
>as
>> media
>> > >>>>>>>>> engines via Emotion modules
>> > >>>>>>>>>   * Selection of media back-end via command-line
>> > >>>>>>>>>   * Album art fetch and caching
>> > >>>>>>>>>   * Video thumbnail timeline generation and caching
>> > >>>>>>>>>   * Works with any Evas engine (OpenGL acceleration, pure
>> software
>> > >>>>>>>>> etc.)
>> > >>>>>>>>>   * Works in X11, Wayland and Framebuffer direct support
>> > >>>>>>>>>   * Accelerated seek on keyboard fowrard/reverse
>> > >>>>>>>>>   * Drag gestures for seeking
>> > >>>>>>>>>   * Special different UI modes for pure audio and video
>> > >>>>>>>>>   * Media center browser/indexer mode
>> > >>>>>>>>>   * Simeline thumbails of videos get generated and
>displayed
>> when
>> > >>>>>>>>> over seek bar
>> > >>>>>>>>>
>> > >>>>>>>>> If you want to see more go to the about page at
>> > >>>>>>>>> https://www.enlightenment.org/p.php?p=about/rage
>> > >>>>>>>>>
>> > >>>>>>>>>
>http://download.enlightenment.org/rel/apps/rage/rage-0.2.0.
>> tar.gz
>> > >>>>>>>>> http://git.enlightenment.org/apps/rage.git
>> > >>>>>>>>>
>> > >>>>>>>>
>> > >>>>>>>>
>> > >>>>>>>> --
>> > >>>>>>>> Massimo Maiurana
>> > >>>>>>>> Ragusa (RG)
>> > >>>>>>>>
>> > >>>>>>>>
>------------------------------------------------------------
>> ------------------
>> > >>>>>>>> _______________________________________________
>> > >>>>>>>> enlightenment-devel mailing list
>> > >>>>>>>> [email protected]
>> > >>>>>>>>
>https://lists.sourceforge.net/lists/listinfo/enlightenment-
>> devel
>> > >>>>>>>>
>> > >>>>>>>
>> > >>>>>>>
>> > >>>>>>
>> > >>>>>>
>> > >>>>>> --
>> > >>>>>> Massimo Maiurana
>> > >>>>>> Ragusa (RG)
>> > >>>>>>
>> > >>>>>
>> > >>>>>
>> > >>>>
>> > >>>>
>> > >>>> --
>> > >>>> Massimo Maiurana
>> > >>>> Ragusa (RG)
>> > >>>
>> > >>>
>> > >>
>> > >>
>> > >> --
>> > >> Massimo Maiurana
>> > >> Ragusa (RG)
>> > >>
>> > >
>> > >
>> >
>> >
>> > --
>> > Massimo Maiurana
>> > Ragusa (RG)
>> >
>>
>>
>> --
>> ------------- Codito, ergo sum - "I code, therefore I am"
>--------------
>> The Rasterman (Carsten Haitzler)    [email protected]
>>
>>
>> ------------------------------------------------------------
>> ------------------
>> _______________________________________________
>> enlightenment-devel mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
>>
>------------------------------------------------------------------------------
>_______________________________________________
>enlightenment-devel mailing list
>[email protected]
>https://lists.sourceforge.net/lists/listinfo/enlightenment-devel

-- 
Inviato dal mio dispositivo Android con K-9 Mail. Perdonate la brevità.
------------------------------------------------------------------------------
_______________________________________________
enlightenment-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/enlightenment-devel

Reply via email to