Skip to main content.
home | support | download

Back to List Archive

Re: coalesce_buffer_max_size

From: Bill Moseley <moseley(at)>
Date: Tue Jul 30 2002 - 15:29:16 GMT
At 08:19 AM 07/30/02 -0700, Peter Asemann wrote:
>Warning: Substituted possible embedded null character(s) in file

Is that a binary file?

>err: Buffer too short in coalesce_word_locations. Increase
>COALESCE_BUFFER_MAX_SIZE in config.h and rebuild.

I think you see that when you are indexing a file that has too many words
(Jose, is that right?) which you would find when trying to index a large
binary file.

There was a case where someone reported that swish was not indexing the
last part of a document.  It turned out that they had an embedded null char
in the doc so swish determined that was the end of the document.  Now swish
reports that warning and replaces the null with a space char.  That allows
swish to continue indexing the document.  But in most cases the warning
means you are trying to index a binary file.

Jose, could COALESCE_BUFFER_MAX_SIZE by dynamically adjusted?

Bill Moseley
Received on Tue Jul 30 15:32:46 2002