On Tue, Feb 21, 2012 at 03:59:34PM +0200, Miika Komu wrote: > On 02/21/2012 03:24 PM, Diego Biurrun wrote: > > review needs-info > > > >On Mon, Feb 20, 2012 at 08:34:18AM +0000, Xin wrote: > >>Xin has proposed merging lp:~hipl-core/hipl/libhip into lp:hipl. > >> > >>Libhip merge proposal: > >> > >>The libhip branch mainly aims to provide a convenient way for system > >>test of HIPL without full installation and configuration. In addition > >>to this purpose, it also provides a socket API alike library approach > >>for application to use HIP. Last by not least, since vanilla Linux TCP > >>does not yet support long periods of disconnectivity [1] , libhip can > >>be a solution in this condition. > >> > >>In the libhip branch, we build a library version of HIP for upper > >>applications, which only exposes traditional socket like API. when > >>using the libhip, hip control messages are transmitted over TCP or > >>UDP, which is similar to TLS/DTLS but we have a unified protocol to > >>handle both datagram and streaming traffic[2]. Compared to TLS/DTLS, > >>this is a big advantage and it may be better use case for HIP[3]. > >> > >>In the libhip, most of code of hipd has been moved to lib/hipdaemon > >>which then becomes a library for both hipd and libhip. By this way, > >>the libhip can reuse the code of hipd to the max extend. This is > >>also the reason why libhip can be a system test approach for hipd, > >>especially in the process of base exchange, the libhip and hipd > >>share the same code base. Meanwhile, the hip daemon, and other > >>functionalities previous exist in the trunk, are kept unchanged and > >>functioning after this merge. > > > >Why is the library called libhipdaemon? > > would something else (e.g. "lib/hipl" and "libhipl") be more suitable? Is it a library for creating hip-related daemon programs? If no, why is there "daemon" in the name? Diego