Re: [NSE] More library global problems

[Patrick Donnelly <batrick () batbytes com>]

On Mon, Feb 8, 2010 at 1:28 PM, Patrik Karlsson <patrik () cqure net> wrote:
> On 8 feb 2010, at 18.09, Patrick Donnelly wrote:
> You should also note in the documentation somewhere that the library
> has static per-thread data. (Usually data like this is kept in some
> sort of state (a table you construct) by the caller that is passed to
> your library's functions (see the html.lua pipeline functions for an
> example). Your approach is probably ok so long as you document it.)
>
> What are the benefits of keeping a state table and passing it to all library
> functions over re-implementing the whole library as a class which is
> instantiated by each script? The request-id would then be implemented as an
> instance variable instead.

Well, if the script needs to start over for some reason, then it can.
Admittedly I haven't attempted to figure out what your library does or
what it's for. In general, scripts should be able to "restart" doing
some operation in a library. The "simplest" method is to discard the
state table and create a new one (from the script writer perspective).

-- 
- Patrick Donnelly
_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev
Archived at http://seclists.org/nmap-dev/