Hi Gilles, Gilles Rayrat wrote: > Ok, so that's what I feared :) you want to re-implement a carob in erlang
Hehe ;) > The problem here is that the protocol is subject to changes... > Actually, the whole point of Carob is to insulate applications from > protocol issues. > Keeping the protocol implementation private brings a great degree of > freedom for > developers, allowing them to easily implement new features, optimize > performance > and even sometimes fix bugs without disrupting backward compatibility. > So, having a third implem (additionally to jdbc driver and carob) will > lead to > yet-another lib to maintain upon any protocol change. > > A better approach could be to make use of Carob via C++ erlang facility: > > <http://www.erlang.se/doc/doc-5.4/doc/tutorial/part_frame.html> I guess you meant this link: http://www.erlang.se/doc/doc-5.4/doc/tutorial/c_portdriver.html My erlang knowledge is not so great so thank you for pointing me towards that page. > You will then get the best of each: an erlang connector much easier to > implement (thanks to carob API) > + the benefit of Carob's updates, sticking to protocol changes, > transparently > > What do you think ? I like the idea, and as I was going to be coding this whilst on *holiday* (timescales on the project are a little tight atm). So I prefer the time expected on this version, but in the long term I will probably implement it in erlang, it works more logically/efficiently for the erlang multiple node context. > This being said, Carob is open source, so if you decide to implement a > native erlang connector, we will be pleased to provide you with the spec. > If you want, you can already have a look at this class javadoc which > contains all commands: > http://sequoia.continuent.org/doc/latest/api/org/continuent/sequoia/common/protocol/Commands.html I'll take a look when I get some time later. Cheers, Dan. _______________________________________________ Carob mailing list [email protected] https://forge.continuent.org/mailman/listinfo/carob
