[CWB] Problem with cqp

[Stefan Evert]

> I should have probably said that this corpus is ca 20% bigger than its previous version. This might be the problem. 1 billion words.

That shouldn't really make a difference, unless your computer is running out of RAM or so (but it should stop with a memory error in this case rather than do weird things).

The usual cause for this kind of problem are very long sentences which lead to buffer overflows in older versions of CQP (I fixed the last bug related to this issue quite recently, so it won't be in the 3.0.0 binary).

> No, I just installed from
> https://sourceforge.net/projects/cwb/files/cwb/cwb-3.0.0/cwb-3.0.0-linux-x86_64.tar.gz/download

That's indeed a bit outdated and has a number of known bugs that have been fixed in the SVN version.


> Please tell me /give me a place to put the corpus and I'll scp it in CQP format to you.

That'll need a lot of bandwidth and disk space, so let's try something else first.  Or could you set up a guest account for me on your machine, so I can take a look there?

> Or should I try the svn first?

Please do.  I recommend the 3.4.x branch even though it's still in beta (and partly alpha) stage; but the latest version in the 3.0 branch should also have all known bugs fixed.

Best,
Stefan