Path: csiph.com!eternal-september.org!feeder.eternal-september.org!mx02.eternal-september.org!.POSTED!not-for-mail From: D Finnigan Newsgroups: comp.sys.apple2.programmer Subject: Re: telnetd for =?UTF-8?B?R05PPw==?= Date: Thu, 7 Jan 2016 22:51:57 -0000 (UTC) Organization: Mac GUI Lines: 94 Message-ID: References: <9a2e912d-963f-44bd-ba8a-5aa4e11ba25a@googlegroups.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Injection-Date: Thu, 7 Jan 2016 22:51:57 -0000 (UTC) Injection-Info: mx02.eternal-september.org; posting-host="ac9e0e4d7dd3957fb97803ee6801bab8"; logging-data="9009"; mail-complaints-to="abuse@eternal-september.org"; posting-account="U2FsdGVkX183TzWCV7PJlhbkKylZ8HUY" User-Agent: Mac GUI Usenet In-Reply-To: Cancel-Lock: sha1:K+dC7LCaC5UiKowaHLSV7hZKRus= Xref: csiph.com comp.sys.apple2.programmer:2065 andrew.roughan wrote: > On Wednesday, January 6, 2016 at 7:47:06 AM UTC+11, D Finnigan wrote: >> Yes, I agree. It would be incredibly shortsighted not to write the >> library >> as a drop-in replacement for the Marinetti tool. >> >> The measure of success is that you delete Marinetti from your system, >> install this new library, and you can use Spectrum Internet Suite and >> SAFE2 >> as if nothing had changed. > > I don't see why a complete replacement of Marinetti would be necessary. > The > Init calls that are currently being made to the TCP, ICMP, IP and UDP > handlers could be extended to check for a Link Layer that knows about > those > protocols and to call the LL directly instead of Marinetti's > implementation. OK, that's one way to do it. > > If there is something that Marinetti does not do that you want it to do > then please raise a Feature Request ticket on Sourceforge. > >> Furthermore, it's an incomplete implementation of some of the protocols, >> such as IP. Did you know that Marinetti has no method for routing IP >> datagrams to either a host on the local link or to a gateway? That's >> because >> when Richard Bennett wrote the original thing in the 1990s, he assumed a >> single connection at the other end of the IIgs. It means that link layers >> such as the LanceGS (the first one) have had to write their own IP >> routing >> algorithm, as well as their own ARP implementation and DHCP client too. > > I have a few issues with with the above, but again, raise a feature > request > if you want an improvement to be made. I believe that's factually correct, and I have issues with it too. Link layer implementers shouldn't have to write all that stuff in their driver. It should be in Marinetti. I found a couple of feature requests that were opened about 10 years ago, that seem similar to what I'm saying. Have these requests been implemented? #16 Common Link Layer for ethernet ARP/DHCP APIs https://sourceforge.net/p/marinetti/feature-requests/16/ #15 Support packet notification from Link Layer https://sourceforge.net/p/marinetti/feature-requests/15/ > >> Who is using other connection methods besides the Uthernet or Uthernet >> II? >> Raise your hands, people. > > Tony Diaz uses the MacIP LL extensively and there are many users of > emulators that use LLs for their emulators. > Agreed. > > With 4MB of RAM as a common installation, lack of RAM is not a problem > that > I have been made aware of. Perhaps not, but who would argue against saving RAM on the Apple II? > > Any 'drop-in replacement' for Marinetti would likely have its own issues > that would need to be ironed out. Perhaps so, but it would be a much smaller code footprint, consisting of a DHCP client, DNS resolver, and wrapper for the W5100's sockets implementation. > > Marinetti is open-source - help make it better. It would be nice if the current source were committed to the public repository on SF. -- ]DF$ The Marina IP stack for Apple II-- http://marina.a2hq.com/