[mythtv] Backend process dies at 4GB file limit?

[Joe Votour]

--- Buzz <buzz at oska.com> wrote:

> Are you interested in this scenario:
> 
> Backend saves files to FAT32 partition.
> Backend tries to exceed 4GB (or there abouts) while
> unattended.
> Backend dies with error "File size limit exceeded"
> emitted by OS.
> 
> 
> Backend's last message prior to dying was:
> "TFW: safe_swite() funky usleep"
> (message comes from ThreadedFileWriter.cpp )
> 
> 
> Buzz.
> 
> P.S. Before you tell me "FAT32 is a bad, naughty,
> unsupported filesystem"..
> tell me another filesystem that works under both MS
> Windows and linux,
> reliably, and I'll happily change (I'm waiting in
> anticipation of native
> ext2/3 windows drivers or NTFS write support in the
> kernel - it's a long
> wait)
> 

<snip>

Unfortunately, there's nothing that the backend can
really do in the current implementation.  The 4GB
limitation is in the FAT32 filesystem itself
(according to Microsoft's documentation).

About all the backend could do is try to split the
file up, but that adds a lot of overhead and
bookkeeping.  However, seeing as there are many Linux
native filesystems that support files larger than 4GB,
I doubt that the developers would be interested in
trying to fix this.

I won't tell you that FAT32 is unsupported (because
that's not true), but it is bad and naughty.  :) 
There are a few ext2 drivers available for Windows,
some free, some commercial.

-- Joe

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com