Hey vincent, Your project looks nice indeed... thinking of this profiling + chart, one thing that comes to my mind is if you could do a chart that nicely handles live data (for life profiling, which is not the case yet), in these scenarios you could always walk in windows and leave some room for more data at the other end (ie: right), new input is only added to the chart and will render quickly unless the window needs to be moved or using auto-scale for axis and the new point triggers a scale change.
overall very good projects On Fri, Jun 30, 2017 at 10:44 AM, Vincent Torri <vincent.to...@gmail.com> wrote: > On Fri, Jun 30, 2017 at 2:42 PM, Stephen Houston <smhousto...@gmail.com> > wrote: >> On Fri, Jun 30, 2017, 5:28 AM Mykyta Biliavskyi <m.biliavs...@samsung.com> >> wrote: >> >>> Hi, >>> >>> On Thu, 2017-06-29 at 08:56 -0300, Gustavo Sverzut Barbieri wrote: >>> > looks very, very nice indeed :-) >>> > >>> > do you have any plans to hook this with system profilers such as >>> > `systemtap`/`oprofile`/`dtrace` and the likes? >>> Interest idea. Right now planned works only with EFL apps. After all >>> core stuff will be finished - it is possible that system profilers will >>> be supported too. I add your mail as ticket into backlog T5642. >>> Thanks. >>> > >>> > On Thu, Jun 29, 2017 at 3:15 AM, Carsten Haitzler <raster@rasterman.c >>> > om> wrote: >>> > > On Fri, 23 Jun 2017 17:11:52 +0300 Mykyta Biliavskyi <m.biliavskyi@ >>> > > samsung.com> >>> > > said: >>> > > >>> > > Well the page is brilliant.. the images/gifs you have... fantastic. >>> Yeah, I like it too, many thanks to Oleksii. >>> > > i tried it >>> > > on an older log file i had and it segfaulted.... :( it seems i now >>> > > no longer >>> > > know how to even produce evlog files now given jackdanielz >>> > > changes... i have to >>> > > figure this out now again... >>> Could you share this log file? Create a ticket on the phabricatorwith log >>> file attached. This is will be a good point to analyze >>> why this happenes and how to fix it inside viewer. We are already >>> checked how viewer works with both types of evlog files: before and >>> > > >>> > > you and jackdanielz need to work together ... i think this should >>> > > go into >>> > > clouseau (with log begin/end controls, dynamic update as logs come >>> > > in and then >>> > > the ability to inspect older logs...). >>> This what I want to talk about, really. I looks into new clouseau >>> client and think that profiling viewer could be a useful part of it. >>> All what we need is a coordinate our actions. Will be great when >>> jackdanielz share his vision accordingly to clouseau tool (maybe a set >>> of tools for debugging ui of EFL apps?). >>> > > and other than the above... >>> > > >>> > > this is BLOODY BRILLIANT. fantastic! like super duper amazing >>> > > fantastic! i >>> > > haven't looked how you do the cpu freq bars (i was using rects... >>> > > basically >>> > > but i found they really did slow things down... so i was tossing up >>> > > doing just >>> > > image objects with custom generated ARGB pixel data). >>> > > >>> The same issue we are had throught development of this version. >>> Hundreds of evas rectangle objects decreases level of satisfaction from >>> using (freezes on redrawing a new time interval). Current solution is >>> drawing (and deleting objects too) was moved into idler time. I was >>> surprised that deleting object is cpu consumption operation too. One of >>> steps to improvement - render a huge images (maybe 8kx10 px) for each >>> cpu core. I think it is natural way for doing this. We are already have >>> this in a backlog to finish it, when time will be available. >>> >> >> I haven't looked at the code and I haven't heard from Cedric recently, but >> it seems like this would be exactly the reason evas_vg was created..... as >> a graphing object. Perhaps it would be something to look into here. > > Like using my chart library using evas vg (see my mail today) ? :) > > Vincent > > >> >>> > this is just simply FANTASTIC stuff. i was planning on coming back >>> > > and adding >>> > > per-frame/window update region info (and display it visually). what >>> > > i had not >>> > > decided on was how to capture the image of the window/frame each >>> > > fram and >>> > > display it nicely... that would make debugging simply amazing to >>> > > have a >>> > > thumbnail or even full image/frame capture per frame as well as >>> > > everything. >>> > > you could see what did and didn't draw that frame (did it overdraw? >>> > > why? what >>> > > triggered that change?). ... go back and inspect "error frames" >>> > > that you saw >>> > > a glitch in but couldn't catch it easily... i hadn't decided on how >>> > > to do this >>> > > with evlog yet... but it'd likely have been a new kind of evlog >>> > > "blob" with >>> > > arbitrary binary data in it not just strings. >>> Screenshots are really adorable feature for this kind of the >>> applications work flows insepcting. >>> After profiling EFL Profiling Viewer I usually received a really huge >>> blob's, like a 400mb for ~2 min. Because a lot of events data was >>> needed for bottleneck investigation. If images will be included into >>> the same blob - it is possible that 1gb size could be reached easly. >>> Mybe we are could add additional event type (for example "@"), that >>> will initiate taking picture. Picture will be stored as usual by given >>> path and name like efl_log_<pid>_<ecore time stamp>.png. And add event >>> record with ecore time stamp data inside blob. And on inspecting time >>> user available to choose folder with images, that are related to >>> choosen evlog binary. >>> > > >>> > > truly awesome work here... nice! >>> > > >>> Also I want to say that Oleksii has added a few really cool features >>> since last time. >>> >>> * Support command line comands. >>> * event double-clicked this event will be zoomed to all grid. >>> * Show events less then MIN_EVENT_SIZE as transparent rectangles. >>> >>> Updated description page: https://phab.enlightenment.org/w/projects/efl >>> _profiling_viewer/ >>> <https://phab.enlightenment.org/w/projects/efl_profiling_viewer/> >>> >>> Todo list: https://phab.enlightenment.org/w/projects/efl_profiling_view >>> er/todo_list/ >>> <https://phab.enlightenment.org/w/projects/efl_profiling_viewer/todo_list/> >>> >>> >>> Many thanks for all your feedbacks and support. It is really important >>> for us. :) >>> >>> > > > Hello all! >>> > > > >>> > > > I am glad to announce that EFL Profiling Viewer is published at >>> > > > enlightenment repos. >>> > > > This project inspired by Raster evlog tool. >>> > > > And done as Oleksii Piskun internship project. >>> > > > >>> > > > Description could be find here: >>> > > > https://phab.enlightenment.org/w/projects/efl_profiling_viewer/ >>> > > > (Please help to place link into Application section on the >>> > > > phriction >>> > > > main page) >>> > > > >>> > > > Source code available here: >>> > > > https://git.enlightenment.org/devs/nikawhite/efl_profiler_viewer. >>> > > > git/ >>> > > > >>> > > > Project and dashboards for ideas and bugs locates here: >>> > > > https://phab.enlightenment.org/project/profile/151/ >>> > > > >>> > > > Thanks for your feedbacks and support. >>> > > > >>> > > > NikaWhite >>> > > > >>> > > > --------------------------------------------------------------- >>> > > > --------------- >>> > > > Check out the vibrant tech community on one of the world's most >>> > > > engaging tech sites, Slashdot.org! http://sdm.link/slashdot >>> > > > _______________________________________________ >>> > > > enlightenment-devel mailing list >>> > > > enlightenment-devel@lists.sourceforge.net >>> > > > https://lists.sourceforge.net/lists/listinfo/enlightenment-devel >>> > > > >>> > > >>> > > >>> > > -- >>> > > ------------- Codito, ergo sum - "I code, therefore I am" ------- >>> > > ------- >>> > > The Rasterman (Carsten Haitzler) ras...@rasterman.com >>> > > >>> > > >>> > > ----------------------------------------------------------------- >>> > > ------------- >>> > > Check out the vibrant tech community on one of the world's most >>> > > engaging tech sites, Slashdot.org! http://sdm.link/slashdot >>> > > _______________________________________________ >>> > > enlightenment-devel mailing list >>> > > enlightenment-devel@lists.sourceforge.net >>> > > https://lists.sourceforge.net/lists/listinfo/enlightenment-devel >>> > >>> > >>> > >>> after jackdanielz changes and all works fine. >>> >>> >>> ------------------------------------------------------------------------------ >>> Check out the vibrant tech community on one of the world's most >>> engaging tech sites, Slashdot.org! http://sdm.link/slashdot >>> _______________________________________________ >>> enlightenment-devel mailing list >>> enlightenment-devel@lists.sourceforge.net >>> https://lists.sourceforge.net/lists/listinfo/enlightenment-devel >>> >> ------------------------------------------------------------------------------ >> Check out the vibrant tech community on one of the world's most >> engaging tech sites, Slashdot.org! http://sdm.link/slashdot >> _______________________________________________ >> enlightenment-devel mailing list >> enlightenment-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/enlightenment-devel > > ------------------------------------------------------------------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > _______________________________________________ > enlightenment-devel mailing list > enlightenment-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- Gustavo Sverzut Barbieri -------------------------------------- Mobile: +55 (16) 99354-9890 ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
