Could you elaborate on why this way is better? (and why we need to install the python2-devel package?)
(or what might be problematic if we just load the python module)
If one (i.e., ROOT) wants to link against “python3” (“python2”) shared libraries, one needs the corresponding “python3-devel” (“python2-devel”) package.
I think one needs the “python” symbolic link only when building ROOT (with “./configure” but not with CMake). Afterward, one can simply explicitly call “python3” (“python2”) and “import ROOT”.
In general, I prefer the system-provided Python versions. I assume they get “bug fixes” at regular intervals, and there are usually plenty of additional packages available (the “numpy” being the most important for ROOT users).
Strangely, you have the 2.7.17 version, though you should have the 2.7.18 (lazy system administrators?).
I don’t trust “extra modules”. They are compiled by somebody at one time. From my experience, they are purely tested, and they never get “bug fixes”.
If you really want a specific version of some software, you need to build it and maintain it yourself (like you do with ROOT).
Hi this is a follow-up: are there any standard checks within ROOT to make sure ROOT (and each module like roofit etc) is built correctly and returns the same result as another ROOT build on other machines? I assume we can use the example macros in the /tutorial directory?
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.