[ID3 Dev] Why is footer not included in total size?

[Jim]

Yes, 2.3 and 2.4 both use synsafe integers for the size in the tag header.
(As does 2.2)

But I was talking about the example that Ben (Allison) gave:

<snip>
> It's easy to imagine, you're done reading the last frame, and then come
> across "3DI[0x04]" and assume it's a frame you don't understand, at which
> point you read the frame size and, woops, now you're somewhere in the
> middle of the file.

Here he is talking about encountering the footer and thinking that it is a
frame.  Then you would read the FRAME size to skip that frame.  I was saying
that was unlikely to happen because (1) 2.3 (or 2.2) parsers shouldn't be
reading 2.4 frames but even if they did try the frame sizes are not stored
in the same format (syncsafe in 2.4 and normal in 2.3) so the parser would
probably be lost long before it got to the footer, if it ever did, and (2)
even a 2.4 parser that does not fully conform should be able to check the
footer flag and subtract 10 from the size given in the TAG header


----- Original Message ----- 
From: "Ray Manning" <ray at homeautomationdev.com>
To: <id3v2 at id3.org>
Sent: Thursday, November 16, 2006 12:54 AM
Subject: RE: [ID3 Dev] Why is footer not included in total size?


>
> I agree that that there is no explicit statement
> in the spec that states a footer cannot appear on a prepended tag.
> The spec only states that padding and footers are mutually exclusive
> which only _implies_ a footer shouldn't appear on a prepended tag
> but since it is not explicitly stated as ill formated, then you
> have to prepare for it.
>
> Both 2.3 and 2.4 use a syncsafe int in the tag header. The 2.3
> spec just doesn't call it a syncsafe int.
>
> You might be thinking of the frame sizes which changed from
> a normal int in 2.3 to a syncsafe int in 2.4 - along with
> the change in the interpretation of the unsynchronisation
> scheme (v2.3 is tag based where v2.4 is frame based).
>
>
> - Ray
>
>
> > -----Original Message-----
> > From: Jim [mailto:jmartin92 at comcast.net]
> > Sent: Wednesday, November 15, 2006 9:23 PM
> > To: id3v2 at id3.org
> > Subject: Re: [ID3 Dev] Why is footer not included in total size?
> >
> >
> > Well, I can see the possibility of the problem there.  However:
> >
> > I know padding and having a footer are exclusive, but since padding is
> > optional for a prepended tag, how does this imply footers should
> > not be used
> > for prepended tags?  (I'm beginning to agree that footers should
> > not be used
> > on prepended tags but relying on something implied by the spec seems to
be
> > asking for trouble.  If footers should only be used on appended tags it
> > would seem better to just update the spec to explicitly say that...and
not
> > leave it up to interpretation that could cause problems.)
> >
> > But let's assume for a moment that a footer CAN be added to a
> > prepended tag.
> > According to the spec, software that encounters a major tag version
beyond
> > what it understands should skip/ignore the entire tag.  (Like a 2.2
parser
> > encountering a 2.3 or higher tag or a 2.3 parser encountering a 2.4 or
> > higher tag, etc.)  How could any software that encounters a tag with a
> > footer do this?  If it uses the size given by the header and skips ahead
> > that many bytes, it would end up exactly at the beginning of the footer
> > instead of after it.
> >
> > I know there is never any guarantee that the size in the header is
correct
> > and will allow you to skip right to the "real" (meaning how it would be
> > without the tag) beginning of the file but here you have a case where a
> > properly written tag with a proper size would land you 10 bytes shy of
end
> > of the tag every time.  That seems wrong.  (Yes, I know that any decent
> > player should check for the sync signal but shouldn't the goal be to
avoid
> > having to scan for this when you have a properly written tag with
> > no "junk"
> > data after it?)
> >
> > The idea that a 2.3 parser would get messed up because it would
> > try to read
> > this as a frame seems remote.  For one thing, as previously stated, a
2.3
> > parser should skip a 2.4 tag entirely.  But if it did try to parse the
> > frames I would think it would already be hopelessly lost since the size
in
> > 2.4 frame headers is stored as a syncsafe integer while in 2.3
> > frames it is
> > not syncsafe.  And for a not fully conforming 2.4 parser it's
> > true that this
> > could be a problem.  But, well, even if it doesn't fully conform to the
> > spec, checking a flag bit and subtracting 10 from the given size
> > is a pretty
> > simple fix.
> >
> > Jim
> >
> > ----- Original Message -----
> > From: "Ben Allison" <benski at winamp.com>
> > To: <id3v2 at id3.org>
> > Sent: Wednesday, November 15, 2006 10:47 PM
> > Subject: Re: [ID3 Dev] Why is footer not included in total size?
> >
> >
> > > The footer is excluded from the length because it would trip up
parsers
> > > that aren't expecting it (either because they are ID3v2.3 or
> > aren't fully
> > > conformant to 2.4).
> > >
> > > It's easy to imagine, you're done reading the last frame, and then
come
> > > across "3DI[0x04]" and assume it's a frame you don't
> > understand, at which
> > > point you read the frame size and, woops, now you're somewhere in the
> > > middle of the file.
> > >
> > > As others have said, footers are meant for appended tags.  Although
not
> > > specifically banned from use on prepended tags, the fact that
> > padding and
> > > the footer are mutually exclusive implies that it's never meant
> > to be used
> > > at the beginning.
> > >
> > > -Ben Allison
> > >
> > > > Hi.
> > > >
> > > > I am wondering what the reasoning was behind the size stored
> > in ID3v2.4
> > > > tags not including the size of the footer (if present).
> > > >
> > > > I've been looking through the source code for some utilities for MP3
> > files
> > > > and some other audio file types.  Many of these utilities use a
simple
> > > > method for skipping over the ID3v2 tag if it is there.  They read
the
> > > > first 10 bytes of the file, do some simple checking (look for 'ID3',
> > check
> > > > that version bytes are less than $FF and in some cases check that
size
> > > > bytes are syncsafe, i.e. < $80) and if it looks like a tag,
> > convert the
> > > > size and skip that many bytes ahead (+ 10 bytes for header).  Most
of
> > > > these utilities would be broken if the tag has a footer.
> > They don't do
> > > > any specific version checking or checking of the flags byte.
> > > >
> > > > If the size stored in ID3v2.4 tags included the footer size (if it
is
> > > > present), it would make these tags much more backwards compatible.
(I
> > > > know it is a pretty easy change to add checking for the
> > footer flag but
> > > > wouldn't it have been better to make it compatible with older
> > apps that
> > > > don't really need to read the tag, just skip it?)   Were there
reasons
> > > > that came up during development why it was done the way it is?  If
so,
> > I'd
> > > > appreciate if someone could share them with me.  Or if there is
> > somewhere
> > > > I can read about this (like some archive of the mailing list) I
would
> > > > appreciate someone pointing me in the right direction.
> > > >
> > > > Thanks.
> > > >
> > > > Jim Martin
> > > >
> > >
> > >
> > > ---------------------------------------------------------------------
> > > To unsubscribe, e-mail: id3v2-unsubscribe at id3.org
> > > For additional commands, e-mail: id3v2-help at id3.org
> > >
> > >
> >
> >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: id3v2-unsubscribe at id3.org
> > For additional commands, e-mail: id3v2-help at id3.org
> >
> >
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: id3v2-unsubscribe at id3.org
> For additional commands, e-mail: id3v2-help at id3.org
>
>



---------------------------------------------------------------------
To unsubscribe, e-mail: id3v2-unsubscribe at id3.org
For additional commands, e-mail: id3v2-help at id3.org