Subject: Re: [ecasound] more potential fixes/features
From: Kai Vehmanen (kai.vehmanen_AT_wakkanet.fi)
Date: Thu Oct 14 2004 - 00:50:57 EEST
On Mon, 11 Oct 2004, Jesse Chappell wrote:
>> The 3-decimals-only limit is there partly on purpose. :) The ECI API has
>> functions eci_command_float_arg() (variant of eci_command()) and
>> eci_last_float() to allow lossless exchange of float parameters. In C++
> Well, any of the C-based ECI api's (most of them) still communicate
> entirely through well-formed-string IPC with the ecasound subprocess,
> and the return values used there still use the same float->string formatting
> I mentioned before. They do not allow lossless float
D'oh! :) Of course, you are right...
> Does this make the configurable decimal formatting option more
> of a high priority? :)
Very much so :) -- will be in 2.3.4.
Hmm, it would be quite straightforward to add a special mode in which
all float return values would be transfered in binary-format over the ECI
protocol. But, but, I guess we'll have to start from increasing precision
of the ascii-representation:
- all ECI parsers would need changes to utilize
the binary-format
- NetECI uses the same protocol format, so a platform
independent binary representation would be needed ->ugh
So ascii it is. If real-life performance bottlenecks are encountered, then
we'll still have the binary-transfer option available.
-- http://www.eca.cx Audio software for Linux!
This archive was generated by hypermail 2b28 : Thu Oct 14 2004 - 00:48:55 EEST