[CWB] CWB: problems with indexing a corpus

[Stephanie Evert]

> On 14 Feb 2023, at 20:08, Mikhail Mikhailov (TAU) <mikhail.mikhailov at tuni.fi> wrote:
> 
> I don't see them anywhere.

They should be in the data directory you specified with the -d flag.

You can try

	cwb-describe-corpus -s <corpus_name>

This should display attributes and their sizes, as well as the data directory of the corpus. Of course, this will only work if you've placed the registry file (with option -R) in the standard registry.

If cwb-describe-corpus is successful, the corpus has been encoded correctly.  It is still not ready for use, though, until it's been indexed with cwb-makeall or cwb-make.

> And when I run the next command:
> 
> cwb-make -V <corpus_name>
> 
> I get this:
> 
> SHELL CMD '/usr/bin/cwb-describe-corpus  -d corpus_name' FAILED:
> >> Non-zero exit value 127.
> >> sh: 1: /usr/bin/cwb-describe-corpus: not found
>  at /usr/local/share/perl/5.30.0/CWB/Encoder.pm line 259.

That means your Perl modules haven't been installed properly.  In particular, they expect a CWB installation in the /usr tree (with command-line tools in /usr/bin), but your actual installation is somewhere else.

If you install CWB from a standard package (.deb / .rpm) and the Perl module via CPAN, this shouldn't happen.

Best,
Stephanie