Some odd things happening with PNG exports on my Mac (10.5, PPC).
I plot distributions with either ROOT or PyROOT, and get the normal default results - black points on white. From the recent trunk of SVN, however, the points on all exported PNG files come out kBlue. This is also the case with JPG files, but not GIFs - which seem to be fine. Examples below.
(As an aside, SVGs never seem to open, and PDFs are not responsive. TIFFs are apparently not supported.)
I tried the same code and data on a SLC4 machine using 5.14/00h (from the Athena 13.0.30 release) - it worked great and did the colors just as I wanted them.
It looks like these files have a huge number of points. The pdf is 23Mb and the SVG 1Gb. GV and Acroread failed to open it. You are right to prefer bitmap output like gif.
Can the “blue effect” be reproduced with any picture or a specific macro is neede on Mac ?
[quote=“couet”]It looks like these files have a huge number of points. The pdf is 23Mb and the SVG 1Gb. GV and Acroread failed to open it. You are right to prefer bitmap output like gif.
Can the “blue effect” be reproduced with any picture or a specific macro is neede on Mac ?[/quote]
This was generated initially with very basic CINT commands. Basically -
TFile *f = new TFile(“BeamHalo.root”)
ntuple.Draw(“pT:X”)
c1.Print(‘file.png’)
Hi,
this problem on PowerPC is known and is related to libAfterImage
(on big endian systems).
I sent report about this to the author (Sasha Vasko).
Hopefully must be investigated and fixed soon.
What do you mean by a “reproducer”? This has been happening only for a week or so now and it happens with both a macro and from the command line. In both cases I can make correct PDF and EPS files, but I also found out that the JPG files also turn out blue.
For reference, this has been happening on an iMac with Catalina OS, but ROOT was working normally since October or so when I first installed it. In light of that, it may be a problem with a recent update, but I don’t know what aspect this is attributed to.
It seems that bug was visible on some specific Mac architecture (PowerPC ?). I was never able to reproduce it. Since 2007 there was no new bug fixes from the author of libAfterImage . May be @Valeriy_Onuchin has more details about it. But as it is 14 years old I doubt there will be anything new.
Hi @darsatdusk ,
as @couet mentioned the real issue is that we are not able to reproduce this. Do you see the issue on LXPLUS, in a Docker container or when installing ROOT from macports, homebrew or conda (see Installing ROOT - ROOT)?