Hi Michael, Roberto and I contacted Alberto De Luca (Bertazza) regarding "if(needFreeRAM == false)" code and this is his answer:
*When I made the changes to the code for reading the rasters I remember having studied RasterImageLayer class for a long time, and some things remained mysterious anyway.* *I think that commenting on the line of code indicated by Michaël eliminates an optimization designed to make the display of small images faster, without recreating them every time by calling the setImage () method.* *By removing that check, the image is recreated every time the users pan / zoom / refresh. I don't think there will be any memory problems, because code that once ran only if there was a lot of free memory (needFreeRAM = false) now always runs. Rather perhaps there will be a decrease in display performance.* *Alberto* Il giorno mer 28 ott 2020 alle ore 15:49 Giuseppe Aruta < giuseppe.ar...@gmail.com> ha scritto: > Hi Michael. > thanks for the comment. I made again the test removing that part of the > code and I had no evidence of that bug (no spaghetti like shape). I was > expecting some out of memory message that I didn't have, maybe because > 37-39 raster layers loaded still use lesser memory that OJ uses on my PC > (currently 6 Gb but I use Ubuntu Mate). > I have no idea why that code was made for. If the reason was to avoid OJ > to use more than 50% of RAM available to draw raster into the view, we can > increase it to 70%. Or to reduce memory consumption on that process. > Peppe > > > Il giorno mar 27 ott 2020 alle ore 23:02 Michaud Michael < > m.michael.mich...@orange.fr> ha scritto: > >> Hi, >> >> Thanks for the video. I also could reroduce the problem. >> >> I reduced the test to its minimum by setting -Xmx100M and trying to load >> the hillshade tif alone. >> >> Surprinsingly, the problem occurs when used memory is about half of the >> available memory. >> >> By chance, I found some code in RasterImageLayer which test when 50% of >> the available memory is used, and managed to eliminate the black and white >> strip effect by removing the test line 390 if(needFreeRAM == false). >> >> But honestly, this is a part of the code I'm not familiar with, and it >> will take some more time if I want to understand why this code change >> eliminate the visual effect and to check if it has unwanted side effects. >> >> The code seems to be from Stefan, and modified by bertazza. If someone >> wants to take the challenge (and explain to me), he's welcome. Otherwise, >> I'll try to study the code and propose a fix within a few days. >> >> Michaël >> >> >> >> As it is related to memory but happens before available memory is >> completely exhausted >> >> *envoyé :* 27 octobre 2020 à 20:00 >> *de :* giuseppe.ar...@gmail.com >> *à :* OpenJump develop and use <jump-pilot-devel@lists.sourceforge.net> >> *objet :* Re: [JPP-Devel] elevation raster and commons imaging >> >> After Roberto's video I was able to reproduce the bug. I had to load 39 >> (!) raster layers together the huge shapefiles (I used the same rasters >> that Roberto kindly provided) before I had that spagetti shape raster our. >> I have no idea about the reason of it. I can image it could be related to >> the delay between the time the Raster is loaded and the time it is >> completely drawn in the view. More files are loaded, more memory is >> engaged, more time it takes to drive display a file. >> Peppe >> >> -- >> Inviato da myMail per Android >> martedì, 27 ottobre 2020, 07:18PM +01:00 da edgar.sol...@web.de: >> >> hey Roberto, >> >> this seems to fall through the cracks ;(. please try and report on the >> following when you find the time >> >> > additionally please try your routine with r6564+ but instead of >> Sextante image use the "Open->File->(Select your tifs, click Next)->(keep >> Use same settings for *.tif enabled, Select *Buffered Image (Commons >> Imaging)*, Click Finished). this merely is a rendering test, raster tools >> will not work on it. please comment on speed, memory usage and overall >> performance incl. bugs. >> >> thx.. ede >> >> >> _______________________________________________ >> Jump-pilot-devel mailing list >> Jump-pilot-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel >> >> >> >> >> _______________________________________________ >> Jump-pilot-devel mailing list >> Jump-pilot-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel >> >>
_______________________________________________ Jump-pilot-devel mailing list Jump-pilot-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel
