Error details in wrappers.

[bhargavvader]

Since all of the wrappers use check_output to run on command line, if an error does occur, there is no information apart from subprocess.CalledProcessError error message. The user must manually run the command on shell again to figure out what's wrong in these cases.

There should ideally be a way for the user to know what the exact nature of the error is, or at least be given instructions on how to figure out what the error is.

[piskvorky]

Good idea!

[bhargavvader]

Would like to add here that having a way to print console log/messages while running wrappers would be very useful.

[piskvorky]

Yes... especially if we can capture that output and direct it to log.

We definitely don't want to clobber our stdout with the subprocess's stdout, unless the user explicitly requests this. IMO logging is the right place for this output.

[kirit93]

To get a verbose error on the command line as you would had the command been run on the shell, the subprocess.check_output function does the trick if used as follows :

If flag is set to true, the subprocess' output will be displayed, and if set to false the subprocess' output will not be displayed.

try:
	subprocess.check_output(cmd, shell=flag)
except:
	pass

Any particular reason a gensim check_output is being used instead of the subprocess method?

[bhargavvader]

@piskvorky , what's the reason we don't just use the subprocess check_output? Or if there was a way we could pass the shell flag in the current function, this would be a very useful fix.

[tmylk]

The reason for custom check_output was support for Python 2.6 which is now removed.
We can use the subprocess.check_output now

[menshikh-iv]

Will be fixed in #1501