I'm asking here as there might ^H^H^H^H^H are people here that know more
than I do about building programs on Unix.
The main question is: Should swish-e be spilt into separate indexing and
Swish-e is commonly used in CGI scripts. Would load times be reduced if
the search program didn't include all the associated code only used for
indexing? Especially since swish-e can be built with libxml2, which is
dynamically loaded at run time on some platforms, and not exactly small
itself, and not needed for searching.
The other issues is the swihs-e C library (perl module) -- and (although
not recommended) -- if the perl module was used, for example, in mod_perl,
where memory is more critical (due to the number of Apache children) would
it be smart to not include the indexing code in that module?
I have no idea how hard it would be to split up swish-e into two programs,
but that's another issue.
[I wish I understood this stuff a bit better. The perl module, for
example, builds against the libswishe.a archive library. If you build a
program (or perl module) against that library does it automatically bring
in *all* the code in that library, and library dependencies? Is that just
the way it works, or does it depend on the dependencies of the modules in
Received on Thu Sep 27 21:40:01 2001