$Id: TODO,v 1.13 2003/08/18 15:56:40 kas Exp $

Unimplemented mrtg.cfg directives
=================================

Htmldir: (it is determined by the .cgi location); no static HTML is generated
NoMib2: (and displaying system uptime) I don't want to query routers from
	the CGI.
Language: TODO
Libadd: (handled separately) Don't want to mess with more BEGIN{} statements.
RouterUptime: I don't want to query the routers from the CGI.
Extension: (has no meaning)
YTics: don't know how to tell this to rrdgraph
YTicsFactor: don't know how to tell this to rrdgraph
Step: don't know what this should do
Option:
	growright: it is the default (don't know how to make growleft graphs)
SetEnv: Makes sense only for Target[]. Not implemented.
Weekformat: Dont know how to do this


Other TODO/missing items
========================

More customizable directory listing.

Exact time of the previous VRULE (for month and year graphs).

The CGI speedup methods other than FastCGI (speedycgi?).

Verify that it works on Windows (I have mixed reports both with
	MS IIS and with Apache on Windows).

Rewrite the config file parsing using MRTG_lib.pm.

Known bugs
==========

* Directory mode does not honour the "Suppress[]: d" and refers to the daily
	graph unconditionally.
* Directory mode does not generate the "WIDHT=" and "HEIGHT=" parameters
	to the <IMG> tag.

* Sometimes it displays 8-times bigger values than MRTG
	(Options[target]: bits usually helps there). Probably it has something
	to do with command-generated stats (Target[...]: `command`) and/or
	stats converted to RRD format from earlier MRTG log databases.

	I was not able to reproduce this, but I have reports of this behaviour.
	I tried to create a fresh mrtg.cfg file with "LogFormat: rrdtool"
	and "Options: bits", and measure the traffic of some switch. Then the
	mrtg-rrd.cgi generated the graphs and values which were in perfect
	correspondence with the real world.

* The Refresh: HTTP header sometimes confuses older browser - when this
	header is added both to the main html file and to the image files,
	the browser sometimes redirects itself to the image URL instead of
	the main HTML page. Netscape 4 and Opera 5 and 6 are reported to have
	this problem (Mozilla, Galeon and MSIE are fine). Since the Refresh:
	header is not defined in the HTTP specification, there is no "correct"
	way to solve this. So I'll leave this unfixed.

* The config file parsing is still not 100% MRTG-compatible, mainly
	with extensive use of [_], [$] and [^] in multiple config files.
	The fix is to use MRTG_lib.pm directly, but I will do it post 0.7
	release.