Fix "bugs" found by static code analysis

[Enrico Bothmann]

Static analysis finds dead code and potential bugs by walking through every possible code path. As some code paths are never used during the run, these bugs mostly are false-positives.†

Still I think it's worthwhile and a concrete measure to improve the code quality/legibility and sometimes even performance/memory efficiency to use these analysis reports.

I dump records here:

​http://physik2.uni-goettingen.de/~ebothmann/sherpa

My suggestion is that each of us looks through and fixes the bugs at least in the module she/he is responsible for, following the discussion at the meeting in Dresden.

• CSShower++/MCatNLO: eno (done)
• PDF: eno (done)
• AMEGIC++: Steffen
• HADRONS++: FrankS
• PHASIC++/Decays: FrankS
• Rest of PHASIC++: Stefan (maybe, cf. his comment)
• METOOLS, COMIX, DIRE: Stefan
• AddOns/Higgs: Stefan
• ...

Maybe you can quickly drop a comment here if you (intent to) clean a module in this way, to prevent work duplication.

† static analysis heeds "assert" statements. E.g. a "dereference of null pointer" bug can be "fixed" by a line like "assert(ptr != NULL)", which also more explicitly states the intent of the developer and leads to more expressive crashes if not met.

[Frank Siegert]

Hi Enrico,

thanks for the code analysis results. I'll take care of HADRONS++ and PHASIC++/Decays.

Cheers, Frank

[anonymous]

Hi Enrico,

I'll take care of METOOLS, COMIX, DIRE and AddOns/Higgs. If noone else volunteers I can do the rest of PHASIC++.

Cheers Stefan

[anonymous]

Hi all, I've uploaded a new report (r27112), we are down from 269 "bugs" to 190:

​http://physik2.uni-goettingen.de/~ebothmann/sherpa/scan-build-r27112-2016-01-13/