[Scrollkeeper-devel] current Scrollkeeper status

[<las...@Su...>]

Hi All,

Dan asked me to write down the current status of the project. 

As far as indexing is concerned Mary sent an email to the list a coupel
of days ago about this.

Everything we have in the CVS now is described in the
scrollkeeper-spec.txt document in the CVS. I updated this a couple of
months ago (my last contribution to Scrollkeeper was at the middle of
July). Since then there was no significant change to the source tree.

Dan has the best ideas about where the project needs to develop from
here.

A couple of small issues and implementation related details that have to
be done are:

1. File locks:

This is about concurrent access to Scrollkeeper databases. In case of
networked Scrollkeeper databases visible for a large number of users and
sysadmins a file locking mechanism has to be implemented during
accessing the databases.

2. popen() calls (security):

There are a couple of popen() calls in Scrollkeeper that are potential
security wholes. The usage of popen() calls is discouraged in code that
wants to be secure.

3. File format improvements: 

This was a long discussed issue, we got to an agreement (it is on the
mailing list), but I never had time to do it. During developing
Scrollkeeper I was asked to have a look at JavaHelp for potential
cooperation between the two systems or just avoiding to reinvent things.
THe JavaHelp metadata formats are in many ways similar to Scrollkeeper,
in some cases superior to that. I suggested changes to the Scrollkeeper
file formats that were accepted, but I never had time to implement them.

4. Relocatable package support:

This is an old issue and there is a solution for it on the mailing list.
Scrollkeeper updates the metadata with the document locations at build
time of the package based on $prefix (in an autoconf and automake
environment). This does not allow relocatable packages.

5. Networked Scrollkeeper database:

I mentioned this before, there is no support for two levels of
Scrollkeeper databases, one for a large number of users (central) and
one for the user only (private). These two merged would give the total
number of documents the user can access.

6. Test plan:

The Sun Gnome team committed to write a test plan for Scrollkeeper and a
test suite in the future. I don't have any idea what is going to happen
with this, a little work has been done in this area only. In case people
want to find out if this is still happening, Leo Binchy
(leo...@su...) should be contacted.

7. OMF directory configuration:

This issue is partly resolved. THe idea is to implement a flexible way
of specifying where to look for OMF metadata files. Right now there are
three ways of doing this, on the command line, through and environment
variable and through a /etc/scrollkeeper.conf file. This is all
implemented. There is one unimplemented idea, namely that at configure
time we should be able to specify through a --omfdirs type of flag the
content of /etc/scrollkeeper.conf. This has been discussed on the
mailing list.

9. Category tree definition:

Very old one, we have an enforced category tree and we had various ideas
about should we enforce a fixed tree on the user or not (discussed a lot
on the mailing list). This will probably need changes once Scrollkeeper
gets tried in real life.

That's all, I hope this helps. Let me know if you have any questions.

Laszlo