Re: [Radmind-devel] design of compression

[Maarten T. <mth...@ci...>]

Wes,

> To maintain the semantics of libsnet, *every* call to snet_write()  
> needs to be treated as the "final" write. (...)

Well, I checked and many packages (like openssl) do a Z_SYNC_FLUSH on  
every write. I guess we can do the same and not cause any interface  
changes at all.

> I'd say that as an initial implementation, we'd skip the function  
> pointers.  Once the second compression method was added, the direct  
> calls can be generalized.  What do you think?

Agreed. Sounds like I can start, the only thing I'm wondering about  
now is the commandline parameters. Use "-g x" on client as well as  
server?

--
maarten