icecube / corsika_reader Goto Github PK

We already have a Corsika namespace so we should remove the Corsika prefix from type names.

We should always use size_t not int or float whenever referring to a quantity of something.
E.g.
int i = auxDepth.size();
int size() const { return int(fN); }

These create compiler warnings that look bad during code review.

For reading data from files, streams are customarily used.

Remove Corsika prefix.

Hello,

The example in /share/examples/cpp/example_1 is not working because there's no ShowerParticleList.h. Here's the error:

fatal error: corsika/ShowerParticleList.h: No such file or directory
compilation terminated.

I'm using the tagged version.

Asserts, used in the tests are currently begin disabled by CMake.

We need to decide on a coding style. Personally it seems strange to put the return type on another line like:
void
CorsikaShowerFile::Close()
Also stuff like indentation should be consistent.

It also seems sloppy to put editor specific code in the source like:
// Configure (x)emacs for this file ...
// Local Variables:
// mode:c++
// compile-command: "make -C .. -k"
// End:

Also "#pragma once" should probably be preferred to:
#ifndef io_CorsikaShower_h
#define io_CorsikaShower_h

Would it be sufficient just to define a few global functions that print to stdout like the following?
log_info(...)
log_warning(...)
log_error(...)

Or maybe would we also want to print the source file and line number?

We should probably try to thing about some common use cases for Corsika Reader and make some examples for these use cases.

Moving this from Trac.

Hi everyone,
i have a technical question concerning the read-in of compressed CORSIKA files (.bz2) for the IceTop? detector simulation. I use the simulation sets 10410 and 10889.
I learned that the current version of the I3CorsikaInjector can not treat bz2 files.
Thus i build the following solution:

1. Copy the compressed CORSIKA file .bz2 file to my directory
2. Unpack it
3. Use it
4. Delete it
   This solution works fine but it is very, very time consuming. Copying & unpacking bz2 files limits the number of jobs i can run in parallel to a very small number because i easily exceed 1TB (i already have some hundreds of GB of data which i really need). I want to run more jobs in parallel, because otherwise i still need months to finish my detector simulation. I do not have this time :(
   The detector simulation i do bases on the detector response of different particle types (separately!) and anyway takes more time than a "usual" simulation.
   Is there a solution? Are there some lines of code i can add in order to make I3CorsikaInjector treat .bz2 files? I heard that it may be possible when "if stream" is changed to "boost::iostreams::filtering_istream", but i am completely unfamiliar with this.
   Can you please help me?
   Cheers, Daniel

Currently we use many different methods of error handling:

1. CorsikaIOException
2. enum Status
3. INFO, ERROR, FATAL

The main question for designing a error handling API is whether we need to recover from and error.
If we need to recover, we should probably use CorsikaIOException and use try{} as needed.
Otherwise we could just use an assert-like method that prints a message and aborts on error.

Check if file is pipe and flag as not seekable if needed.
Something like:
S_ISFIFO(s.st_mode) || S_ISSOCK(s.st_mode)

The simple example code

raw = corsika.RawStream(filename)
block = corsika.Block()
raw.get_next_block(block)
print(block.ID)
raw.get_next_block(block)
print(block.ID)
particles = raw.particles()

for p in particles:
      print(p)

is causing an error message for the particle iterator

RUNH
EVTH
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Input In [4], in <cell line: 15>()
     11 print(block.ID)
     13 particles = raw.particles()
---> 15 for p in particles:
     16       print(p)

TypeError: iter() returned non-iterator of type 'RawParticleIterator'

Same error for ShowerFileParticleIterator. Most likely due to python3

As far as I can tell RawFile is simply a wrapper around VRawStream.
Is there any reason to keep it? @javierggt

Currently the file is opened an extra time to check the format spec.
Maybe copy this data to the first block instead?