[sword-devel] Web Interface

Derek Neighbors sword-devel@crosswire.org
Mon, 24 Feb 2003 07:10:24 -0700 (MST)

Chris Little said:
> Though I agree portability should be a goal, I think these are actually
> minor concerns.  There aren't going to be and don't need to be throngs

Perhaps, I didnt put in good terms.  The more you deviate from the Free
Software user base the less help you get.  The less help you get the more
your product suffers.  Let's say for example I pick a license that is
'incompatable' with the GNU GPL for a free software product.  In essence,
I am locking out tens of thousands of packages and their code from being
reused within my application before I write a single line of code.  This
means I am at the disadvantage of writing everything myself.  In the same
way choosing a tool that is not widely accepted in the Free Software
community means doing a lot of work yourself.  Java while it has some free
implementations is not widely accepted because of the licensing of Java

> of servers providing the same identical interface.  There will be some
> since some people like to control their whole site, but we don't need to
> target people who have 10 meg user accounts that come free with their
> monthly service or even most people with paid hosting plans as potential
> users of the server-side componenets.  If you don't have the ability to
> install your own software on the server (like libraries or Tomcat), you
> probably ought to just link to someone else's site.

I own a hosting company and have full control of my servers, but no way I
would consider installing stuff like Tomcat, because of the licensing of
Java.  I can surely link to other peoples sites, but why burden them with
my traffic, if I can take it on myself?  I assume you have read peer 2
peer theories and how it affects information exchange?  It seems you
disregard studies in this area and believe that a 'few' mega hosts would
better serve the people than lots of distributed hosts?

> If Troy is going to do the implementation, I think the options are Java,
>  C, and C++.  He seems to have religious objections to Perl and PHP. :)

Certainly if Troy is doing all the code, he can do it in whatever he
prefers.  I pretty much abhor Perl and PHP as well, but I suggest them
because they are widely used, have lots of existing libraries and modules
available for reuse that relate to web applications and would allow this
project to garner the largest number of contributors.  Personally I prefer
python, but notice I'm not suggesting it.

>  Since JSword is not at the level needed (most of the module drivers and
>  filters still being absent I believe) I think you can probably expect
> something in C/C++.  It's just a guess, but I'd bet money on it (except
> that now Troy would implement in something else just to make me lose I'm
>  sure).  So I think portability issues won't be any problem.

C/C++ is good.  You wont get as many web monkey's to 'help' out and you
lose a lot in the way of existing web type libraries, but at least you can
allow anyone with cgi support replicate functionality to distribute the
workload which is a plus.