[sword-devel] Sword ideas

Chris Little sword-devel@crosswire.org
Sat, 20 May 2000 14:00:04 -0700

I'm not too sure whether we want to gzip our entire data files.  KWord,
XMMS, Netscape, etc. can all do this because their data files are read in
whole.  i.e. they load the entire file into memory.  We don't normally do
this with sword.  Instead, portions of the file are looked up in the index
tables and read from the data files.  For example, if I wanted John 3:16
from the KJV module, we only need to read John 3:16 into memory.  If we
gzipped the whole KJV module, it would have to be uncompressed into memory
first before we can retrieve the particular verse we want.

On the topic of compressed modules, I think we have something in the works
or possibly completed that is more suited to us.  Can anyone clue in those
of us who've been out of the loop for a while (i.e. me) about the compressed
module support?

On the topic of easing installation be making our modules consist of a
single file, that might not be a bad idea.  We could build in tarball
support without affecting the indexing except by constant offsets that could
be read from the tarball.  I'm not sure if we need to worry about easing
installation in this way, however.  It might be more effective to simply
push people to use the installers.

About using a uniform format, I have to disagree.  I think we should use the
format that best suits the text.  For Bibles with markup, that is typically
GBF.  For other texts, it's typically ThML (but someone's being lazy and
hasn't finished the filters, sorry -- unless some other kind sould took that
over from me :).  Often enough, though, plain text is the best choice.
Having lots of supported formats available gives us versatility.  What
aspects do you perceive it as over-complicating?  Does it make front-end
authoring too complicated or is it something to do with module authoring or
user installation?