[physfs] Buffering API...

[Edward Rudd]

Cool, I'll have to check that out..
Currently I have a C++ wrapper around PhysFS derived from the IOStreams
classes and am using Buffering via the stdc++ iostream.  I've attached
that to this message. I'd like some opinions and comments on how I
implemented the main class interface.. 

This is currently distributed "as is".  I'd like a credit to me if you
use it.. (as in Using Code by Edward Rudd <eddie at omegaware.com>)  


On Sun, 2002-12-01 at 05:32, Ryan C. Gordon wrote:
> PhysicsFS, in the latest CVS, now has the ability to buffer i/o to memory.
> 
> To use it:
> 
>   PHYSFS_file *f = PHYSFS_openRead("/my/dir/myfile.ext");
>   if (f != NULL)
>   {
>       int rc = PHYSFS_setBuffer(f, 2048);
>       // if rc != 0, you've got a 2k buffer for that file now.
>   }
> 
>   Once a file is buffered, the next call to PHYSFS_read() will fill the
> buffer, and will take data from there until it needs refilling.
> 
> If you buffer a file opened for write/append, all calls to PHYSFS_write()
> will fill that buffer until it's full, and then flush it to begin
> refilling. If you want to explicitly flush it, you call:
> 
>     PHYSFS_flush(f);
> 
> AND CHECK THE RETURN VALUE! Since we're buffering, PHYSFS_write() could
> succeed but the flush (or close, which implicitly flushes) could fail.
> 
> If you want to unbuffer a file, call PHYSFS_setBuffer() with a zero. Files
> are unbuffered by default (assuming the OS allows this).
> 
> 
> seek, eof, tell, etc all take the buffer into account and should work
> transparently.
> 
> There are commands in test_physfs for testing the buffering code:
> 
>    "setbuffer x" will set a buffer on future "cat" and "append" and
> "write" commands and "stressbuffer x" will run a stress test (please set
> the writedir and searchpath to "." first).
> 
> This is only lightly tested, but it seems fairly functional. Please pound
> on it. I'll package up physfs 0.1.8 in a day or two.
> 
> --ryan.
> 
> 
-- 
Edward Rudd <eddie at omegaware.com>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: PhysFS-0.1.tar.gz
Type: application/x-gzip
Size: 153806 bytes
Desc: not available
URL: <http://icculus.org/pipermail/physfs/attachments/20021202/ee749838/attachment.gz>