Hi Rooters,
I have used Axel Naumann’s visual c++/cygwin pages (muenster.de/~naumana/rootmsvc.html) to build 5.11.06 with Visual C++ 8.0 (2005) and Bertrand Bellenot’s MFCRoot application runs well.
I had to:
Change the vc++ target from “c:\root\bin” to c:\root\bin"
Rename c:\cygwin\bin\link.exe (not user\bin) to link_.exe.
Remove libEGPythia6.lib from the vc++ library list.
In my app, on the second pad of a canvas with twelve divisions
the call
fTree->Draw(cFormat,cChanCut,"");
creates an access violation at the following line in
ttreeplayer
// invoke the selector
Long64_t nrows = Process(fSelector,option,nentries,firstentry);
I do not think libEGPythia6 is a problem. Also, I am not in a
hurry to resolve this issue (the msi binary for 5.11.06 works).
Still, all suggestions are welcome.
Axel,
I stated things very badly: The first ->Draw call works - it is on a second pass through that the it fails. As soon as I can I will try to understand the difference. (Attached is the vc++ debugger expansion for the “good” tree and the “bad” tree structures - they are almost identical.)
Having stopped the second call I can see all of the plots but when the program exits 5.11.07 also has memory leaks that don’t appear with the .06 msi version. Later this week I will update from the CVS and rebuild.
Cheers,
Steve bad-draw.txt (5.73 KB) good-draw.txt (5.73 KB)
Hi Rooters,
More hints - I can stop the memory exception by not running the draw command in batch mode. This doesn’t appear to save me time anyway. As for the memory leak it is also there with Bertrand’s project. I will check back mid-week.
Cheers,
Steve
Axel,
I have untarred the 5.12.00 sources, compiled them, and run both applications successfully - no exceptions, no leaks. Thanks for your very helpful instructions. By the way, I am running without a console and Ttree->print() shows nothing, but it is now moot.
Cheers,
Steve