[CELT-dev] CELT grabbing 100KB of memory right off the top

[Brett Paterson]

there's no readme.win32 in the snapshot tarball.

Brett Paterson | CEO
FMOD by Firelight Technologies Pty Ltd
Interactive Audio Middleware | www.fmod.org
PH: +61 3 96635947 Fax: +61 3 96635951


-----Original Message-----
From: Gregory Maxwell [mailto:gmaxwell at gmail.com] 
Sent: Tuesday, 19 April 2011 2:55 PM
To: brett at fmod.org
Cc: Jean-Marc Valin; celt-dev at xiph.org
Subject: Re: [CELT-dev] CELT grabbing 100KB of memory right off the top

On Tue, Apr 19, 2011 at 12:32 AM, Brett Paterson <brett at fmod.org> wrote:
> We don't even use configure, you are expecting us all to install mingw on
> windows or something just to set this up?  What about embedded platforms
> like consoles?

I don't believe anyone was trying to suggest using configure.

The expected procedure is that anyone building on an embedded system
will need to build a setup header file for their system.  README.Win32
recommend setting this in config.h and then -DHAVE_CONFIG_H in the
build options, though it appears that we should make the instructions
more prominent.

We've put in considerable effort to portability and we'd certainly
like to improve it.

No two embedded systems are alike— and the 'portable' build tools like
automake are not all that portable when it comes to embedded systems
as you well know, so some configuration is required. Libcelt runs on a
variety of embedded systems, including ones where int is 16 bits and
ones where char is 16 bits, so some configuration is absolutely
required. It would probably be helpful for us to collect a set of
pre-made files for this purpose.

As far as the define namespacing goes— we're talking about
pre-processor macros which only exist in the libcelt code. They
shouldn't be visible elsewhere, and we're only using them to NOP C99
reserved keywords.  I'm sorry, but I'm still not seeing the practical
issue on that point.