|Laura Sheble wrote:
> We have been trying to install 2.39 on Solaris 8 with some problems. We
> found a lot of help from comments by Don Gourley (see below - Notably, we
> have not yet linked the XML parser into INC - if this is necessary, help on
> how to do this would be appreciated) and others.
> We still have some problems. Here are the symptoms:
> 1a. Though the demo collection and administrative options appear on our site
> at gsdl/cgi-bin/library, the demo collection appears to be empty. Searches
> retrieve no results & there are no options to browse. The option to "Search
> for _stx_ for ..." is available.
> warning: gdbm database "/opt/local/gsdl/collect/demo/index/text/demo.bdb"
> does not exist
This sounds like the collection has not been successfully built.
_stx_ and _dtx_ etc are the default names for the section text and
document text indexes, but if the collection is built these names are
overridden with more sensible names, such as "sections" and "documents".
The gdbm file must exist for a collection to be successfully built.
Greenstone distributes the "demo" collection unbuilt.
> 1c. I also noted that there is no import directory in the demo directory,
> though there are many files (png, others in HASH.dir & HASHNULL.dir in the
> archives directory for the collection).
> 2. Tried to build a collection from the command line using:
> Ran perl -S mkcol.pl -creator email@example.com downt
> received the error message:
> Can't find mkcol.pl on PATH
You need to set up the environment first for greenstone to find its
files. $ source setup.bash for bash/ksh/sh-type shells, or
$ source setup.csh for csh/tcsh shells.
mkcol.pl creates a brand-new collection. You probably want to run
import.pl to make the archives, and then buildcol.pl to create the
indexes from the archives. You might need to copy the import directory
again, as the Collector web front-end removes the import directory after
creating the archives directory.
> 3. Despite the problems with the demo collection & creating a collection
> b. while clicking on a link from the contents section results in the error
> External Link
> The link you have selected is external to any of your currently selected
> collections. If you still wish to view this link and your browser has access
> to the Web, you can go forward to this page; otherwise use your browsers
> "back" button to return to the previous document.
> clicking on a link from the menu results in the error:
> Internal Link not Found
> For reasons beyond our control, the internal link you have selected does not
> exist. This is probably due to an error in the source collection. Use your
> browsers "back" button to return to the previous document.
This is because the HTMLPlug tries to resolve links between pages if
both pages are imported, as well as letting people know if they are
following a link outside of the collection. If you don't want this
behaviour, add the "-nolinks" option to HTMLPlug in the collect.cfg
file. This will break relative links between your pages though.
Hope this helps