Re: [RFC PATCH 1/1] Define an extended tree format

[Junio C Hamano <gitster@pobox.com>]

"brian m. carlson" <sandals@crustytoothpaste.net> writes:

> +The first part of what is traditionally the file name consists of a modified
> +BER-encoded integer representing the number of bytes in the extended
> +information section not including this length itself; that is, the number of
> +bytes which must be skipped to reach the file name.  This allows parsing
> +unknown values in a graceful way.

In short, you prepend extra bytes that contain no NUL in front of
the pathname part of the tree entry, but that extra bytes begins
with the length to allow skipping.  

The tree entries must be in sorted order, but I presume that this
extra information merely happens to be encoded inside pathname
field, but does not contribute to the ordering of the entries?

For debuggability, I am not sure if modified BER is the best format
(it is essentially binary gibberish).

For extensibility, I wonder if we should allow more than (type,
flags) to be given in the future?  I guess a new <type> will allow
extra things after <flags> (like conflict tree takes stage number),
so the format is already extensible (we do not want to see people
willy-nilly add extended stuff to the tree object anyway, since it
would allow the logically same object represented by multiple byte
streams that are not bit-for-bit identical, so extensibility is a
double-edged sword, but still necessary).

When we have more than one stage #1 entries (can happen in a
criss-cross merge) or multiple stage #3 entries (octopus), what sort
order do we apply to them for conflict tree entries?

As you said, it does look ugly ;-)