Re: IP Management Tool - Minimum Requirements
- Date: Sat, 21 Apr 2001 18:59:55 +0200 (CEST)
On Thu, 19 Apr 2001, Michael van Elst wrote:
> Why not ? Most assignments are completely arbitrary. There are two
> reasons to chose a specific subnet: aggregation and reservation.
> In both cases you do a bigger assignment first (which only exists
> privately, not in the RIPE database) and select your assignments
> from this block.
Implementation suggestions that spring to mind:
1. since both perl and PHP support object techniques, make the subnet
selection strategy a class (or module, or whatever the language-specific
term of the week is). Instantiate a configured class to switch easily
between strategies. This probably generalises for other program areas.
2. this system may have multiple users accessing it simultaneously. subnet
selections etc cannot be allowed to conflict. PostgreSQL has far more
mature support for the proper transactions that are motivated, and also
has a native IP address/subnet datatype with corresponding operations.
3. multiple front-end support should be possible. I can guarantee that
some LIRs will want to customise it, possibly drastically. Again, an
object class (module) can support this, or you can just spit out XML and
have XSLT parse it for HTML/pdf/etc - or use both strategies. This is
surprisingly easy.
4. this is off-topic for the lir-wg and needs a dedicated mailing list!
see [7]
5. once a dedicated list is up, a code license has to be chosen, and then
a cvs/build server provided. I suggest these issues wait, but they are
of equal importance with the choice of implementation language and
deployment environment.
6. php can be used outside of a web interface, but perl is still more
developer-friendly in the long run, has a wider range of libraries,
greater code maturity, a larger userbase especially in this
community, and proper structuring will allow development of Perl/TK
interfaces to logic core without interference with web FE's - et
cetera.
Complicated data structures in Perl are *not* a pain - in fact,
they're considerably more powerful, expressive and concise than those
available in the language cores of all other language candidates,
as anyone who's used the Schwartzian Transform will agree.
7. this is off-topic for the lir-wg and needs a dedicated mailing list!
see [4]
- Joshua
