[FFmpeg-devel] [PATCH] Add av_file_get_size() and av_file_read(), replace cmdutils.h:read_file().

[Michael Niedermayer]

On Mon, Dec 06, 2010 at 12:02:18AM +0100, Stefano Sabatini wrote:
> On date Friday 2010-12-03 18:43:46 +0100, Michael Niedermayer encoded:
> > On Fri, Dec 03, 2010 at 05:15:44PM +0100, Stefano Sabatini wrote:
> > > On date Thursday 2010-12-02 02:32:30 +0100, Michael Niedermayer encoded:
> > > [...]
> > > > > +/**
> > > > > + * Read the file with name filename, and put its content in a newly
> > > > > + * allocated 0-terminated buffer. If filename references a link, the
> > > > > + * content of the linked file is read.
> > > > > + *
> > > > > + * @param bufptr location where pointer to buffer is returned
> > > > > + * @param size   location where size of buffer is returned
> > > > > + * @param log_ctx context used for logging
> > > > > + * @return 0 in case of success, a negative value corresponding to an
> > > > > + * AVERROR error code in case of failure
> > > > > + */
> > > > > +int av_file_read(const char *filename, char **bufptr, size_t *size, void *log_ctx);
> > > > 
> > > > this API is crap.
> > > > at a minimunm it should support memory mapped files on platforms supporting them
> > > 
> > > We could have a:
> > > int av_file_get(const char *filename, size_t *size, void *log_ctx);
> > > 
> > > (alternative name: av_file_open())
> > > 
> > > which returns the filedes and the size performing the boring checks,
> > > then the application may suck the content in a buffer or using mmap()
> > > access. Would be this acceptable?
> > 
> > its worse in every respect, and still doenst support mmap
> > 
> > what i was thinking of:
> > @param **bufptr The file content, can be a allocated buffer or access through mmap
> > fd=av_file_map(const char *filename, uint8_t **bufptr, size_t *size, int access_rights, void *log_ctx);
> > 
> > av_file_unmap(int fd, uint8_t *bufptr, size_t size);
> 
> Yet this API is overkill and awkward when you only need to open a
> file, allocate a buffer and copy the file content to it, without to
> keep track of the fd and size (the buffer may need to be released
> later in a different context), as it is the case for ffmpeg.c.

You need to keep track of these things in your API as well. (fd to close(),
size, to not read/write over the array end)
but your API is not suitable for anything but your use case.
One could give the functions a struct context, but if that would be simpler
iam not so sure.

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

The real ebay dictionary, page 2
"100% positive feedback" - "All either got their money back or didnt complain"
"Best seller ever, very honest" - "Seller refunded buyer after failed scam"
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20101206/2299bc8f/attachment.pgp>