> > Zero-length data can't be used to indicate that a file is a Freenet
> > control file in my scheme anyway because the non-metadata portion of the
> > file should be contained in the data section of the file.
> After your argument that you wanted a clean protocol, you go and do
> this?  Now *thats* dumb.  Why split it up this way?  Is the client going
> to care when its parsing the FNP message?  It doesn't have this sort of
> crossover in the wire protocol, because its totally unnecessary.
> 
> You're sticking too hard and fast to the names 'metadata' and 'data'.  We
> have two sections in an encrypted partition.  In the case of a normal
> document we call them 'metadata' and 'data'.  In the case of a
> freenet-special file, this doesn't really make much sense.  I'm just
> trying to cover all the endcases with the most natural solution:

Why doesn't this make sense? Special Freenet files have data just like
normal files. An index is certainly just like a normal file. In fact
indices currently *are* normal files. I don't see why we should make a
special distinction between the special freenet files and other files in
terms of format. The only difference between them is that you need a
Freenet client in order to make use of a special Freenet file and most
special Freenet files will be supported by most clients. Other than that,
they can be represented just like normal files and I don't see why it's
advantageous to represent them otherwise.

> Yeah, but putting data in the datafield on a freenet-special message is
> dumb.

Why? How is it better to mix the data with the metadata?

Example message:
Index
Author=blah
Title=blah
Signature=blah
Entry.1=freenet:CHK at q23123123123
Entry.2=freenet:CHK at jkljklfg8ugf879gf87
Entry.3=freenet:CHK at 548945894589045890
Data
<EOF>

Or:
Author=blah
Title=blah
Signature=blah
Content-type=freenet/index
Data
freenet:CHK at q23123123123
freenet:CHK at jkljklfg8ugf879gf87
freenet:CHK at 548945894589045890
[EOF]

Compare to normal file:
Author=blah
Title=blah
Signature=blah
Content-type=image/gif
Data
<some gif data>
[EOF]

Why is the first way better than the second? The second is consistent with
the normal way to represent a file, moves the freenet special file
handling code into the same modules as the other file handling code which
most clients will have to handle files of different types avoids the
possibility of namespace collisions in the metadata section.



_______________________________________________
Freenet-dev mailing list
Freenet-dev at lists.sourceforge.net
http://lists.sourceforge.net/mailman/listinfo/freenet-dev

Reply via email to