Re: [reSIProcate] Uri.cxx and encoding
Dale R. Worley wrote:
On Thu, 2006-02-23 at 15:23 +0000, Daniel Pocock wrote:
b) compatibility - some other SIP implementations send the # symbol
un-encoded (Asterisk), and expect to receive this symbol un-encoded (we
have found one case of this with someone using a Nextone/Cisco combination)
See
http://list.sipfoundry.org/archive/resiprocate-devel/msg04248.html
OK, I'm starting to remember now.
From a theoretical point of view, I would argue that any change which
allows Resip to send invalid URIs should be avoided at all costs. But
we may be stuck having to deal with this as a compatibility issue. It's
not clear that it's worth expending energy to make such horrible hacks
*efficient*.
It's not about making the hack efficient - it actually makes the whole
Uri processing operate more efficiently. The hack for sending # is just
a bonus.
The behaviour is only enabled at the discretion of the application
developer, who could achieve the same thing by hacking up the code
himself. I believe my comments in the code do provide some hint that
these methods make the Uris non RFC 2396 compliant, so hopefully they
won't be used excessively or by mistake.
As a meta-issue, remember that when you post code, your readers may not
remember why the code was written, and you would probably get better
responses if you remind them.
That's a good point, VoIP is a fast moving area of open source
development and I'm sure there are probably a few new people on the list
in the weeks since I made the original post.