STL question...

[Thomas H. Ptacek]

Click is amazing; I've been watching it for many months now, and
using it for expiriments on our test networks. The fine-grained element
forwarding model (and, generically, the notion of "router plugins") has
radically influenced my designs.

I'm working on a commercial project to do application-layer multicasting
using packet forwarding as the primitive (rather than "flow" or "stream"
switching/splitting). We've already finished and "shipped" our first
few revs of the software, which is modelled after and scales like a web
server with no backing store (it's all userland UDP/TCP packet/record
switching). We're very successfully using a stripped-down variant of
OSPF to build distribution trees with some very naieve multicast routing
algorithms (which work nicely in application-layer routing!).

In any case, my thought was to start modelling the packet forwarding of
this system in Click; my only issue at this point is that my codebase
is all STL, and Click seems to studiously avoid STL data structures.
I'm curious if there's a real reason why you don't use Standard C++ 
strings and vectors (the hash table library has been useful to me, though,
thanks!); you guys started in '99, so maybe there was an STLport
stability issue then? Or is there a more subtle problem I'm missing?

I'm dangerously close to backing your data structure implementations with
STL. =)

Thanks again for some incredibly cool software.

---
Thomas H. Ptacek
Sonicity, Inc.