
This document describes repository index information file format and
its processing conventions. Repository index information file is
stored in `base' directory and has name `info'. It is the first file
to read during repository index loading.

This file consists of set of pairs like `name=value'. Name cannot
include `=' character, any number of spaces is allowed around `=' sign
between name and value. Empty lines are silently skipped, any part of
line after `#' character without previously used `\' character is also
skipped as comment. All spaces and tabs at the beginning or end of
value string are removed. Parameter values must be processed in case
sensitive mode.

The file has the following meaningful parameters (may be extended
later):

* format_type;
* format_version;
* compression_type;
* md5sum_file.

Repository index may be saved in text or binary format. Text format is
well suitable for debugging purposes but binary format takes less disk
space. So `format_type' may have two values: `binary' and
`text'. `format_version' parameter stores minimal required version of
the package to be able to read info file properly.

Parameter `compression_type' saves information about used compression
library. The following values can be used: `none', `gzip', `bzip2' or
`xz'. Of course, not all of them are currently supported. It is just
declaration of possible values. `md5sum_file' stores the name of
possibly signed md5sum file to control integrity of repository index
files.

Parameters listed above are strongly required. Information file must
be considered invalid if some of them are absent. All other parameters
with name constructed with small and capital English letters, digits
and `_' character are allowed for any additional user information and
must be skipped during file reading process.
