alfs Requirements [Was blank]

Bruce Dubbs bruce.dubbs at gmail.com
Mon Oct 20 14:02:05 PDT 2008


Jeremy Huntwork wrote:
> Bruce Dubbs wrote:
>> I'll have to disagree here.  I prefer an interpreted language like bash, python, 
>> or php.  I don't like perl because I think the syntax is error prone.  The ease 
>> of viewing, understanding, testing, and debugging an interpreted package is the 
>> dominant issue.
> 
> It's a factor, but I don't know if I'd call it a dominant issue. Poring 
> through someone else's bash scripts can sometimes be just as daunting as 
> C or C++.
> 
>> The speed is a non-issue as the compile times in LFS dwarf any alfs execution times.
> 
> In turn, I disagree with this statement (at least the part about speed 
> being a non-issue). When I'm in the development process and am 
> constantly adjusting things and re-parsing the book, 1 second compared 
> to 20 seconds makes a difference.
> 
> In any case, my biggest complaint with the way jhalfs currently parses 
> the book has more to do with its reliance on libxslt and trying to 
> understand the stylesheets that make up the logic behind it.

If you are going to reinvent the ability to parse, then good luck.  I've done 
<understatement>just a bit of C/C++ programming</understatement> in my career 
and I don't think you will be able to parse the book in a second unless you come 
up with an algorithm that is quite different from xsltproc.  To do that reliably 
will take many times 20 seconds.

BTW, understanding the stylesheets is easier if you just think in stacks.  It's 
really not a lot different from Lisp.

   -- Bruce



More information about the alfs-discuss mailing list