Actually tell why file read fails #51
Conversation
ykempf
left a comment
ykempf
left a comment
There was a problem hiding this comment.
While at it I suppose also the other MPI_* calls like closing could be improved?
vlsv_reader_parallel.cpp
Outdated
| parallelFileOpen = true; | ||
| } else { | ||
| success = false; | ||
| cerr << "Failed to open parallel file with MPI error " << error << endl; |
There was a problem hiding this comment.
Is this now a number, or some string?
There was a problem hiding this comment.
That's an int. I suppose the proper way would be to have a function mapping error codes to descriptions of MPI errors but now it at least doesn't just gobble up the error code and spit out the non-answer of "there was an error".
There was a problem hiding this comment.
There was a problem hiding this comment.
Yeah that tracks, just needs a whopper to convert that from C-string to something sensible.
There was a problem hiding this comment.
Can't that be plonked to cerr? Or to a stringstreamstream and then that gets cerr'ed?
There was a problem hiding this comment.
Look at the function signature, the actual string "returned" is an argument. So the conversion is sth like:
char c[MPI_MAX_ERROR_STRING];
int len;
MPI_Error_string(err, c, &len);
std::string s (s, len);
| else parallelFileOpen = true; | ||
|
|
||
| if (success == false) cerr << "Failed to open parallel file" << endl; | ||
| int err {MPI_File_open(comm,const_cast<char*>(fileName.c_str()),accessMode,mpiInfo,&filePtr)}; |
There was a problem hiding this comment.
"While valid, this is stupid." :) Initializers do have their uses but let's be realistic, here it's just odd. (TBH is odd also on line 420 since it's just a single variable)
3 participants
No description provided.