0.99.31
From: Tobi
0) hmpf ... forgot to add rrdresize.pod and friends to MANIFEST
 
0.99.30
From: Alex
0) We can now alter the number of rows in an RRA with rrd_resize

0.99.29
From: Blair
0) Prevent Makefile from rebuilding the perl modules on every run

0.99.28
From: Blair
0) fixed filedescriptor leak in rrd_graph
From: Alan Lichty <alan_lichty@eli.net>
0) fixes for rrdintroduction
1) contrib/log2rrd for mrtg *.log file to *.rrd conversion
From: Tobi
0) fixed data_fetch bug ... I did not set the requested
   data resolution properly ... the data resolution returned
   by rrd_fetch was more or less random ...

0.99.27
From: Tobi
0) fixed the - file-name to really work.
1) The rrdintro is now in the doc tree.

0.99.26
From: Tobi and Jeremy
0) fixed documentaion GNUMakefile
From: Kai Siering <kai.siering@mediaways.net>
0) Updated rrdgraph.pod ...
From: Tobi with inspiration from Kai
0) if rrdgraph get - as a filename, its output will go to standard out
1) ****INCOMPATIBLE***** the IF operator was not taking its argments in the
  correct order ... I have put an example into the manpage to show how
  it should be done and how it is now done ...
From: Alex van den Bogaerdt <alex@ergens.op.het.net>
0) Contributed two tutorial, make sure you visit
   bin_dec_hex.man and rrdtutorial.man in the doc directory.
 
0.99.25
From: Blair Zajac <bzajac@geostaff.com>
0) do not set PATH in top Makefile ... 
From: Tobi
0) added better IEEE test to configure

0.99.24
From: Jeremy Fischer <jeremy@kiva.net>
0) install functionality for Makefile
From: Tobi
0) removed TOTAL  function ...

0.99.23
* fixed cache bug in configure script ... CFLAGS_EXTRA were disregarded in
  the second run ... found by Douglas Brunk

0.99.22
* From Russ Wright <rwwright@home.com>
0) use PIC instead of pic
* From Tobi
0) fixed a memory leak in rrd_graph (p_data was not fully freed)
   purify is a cool tool indeed :-)

0.99.21
* From Tobi
0) Checking for IEEE math in configure ... find out if -ieee helps
1) rrdtool checks its base arguments with strcmp
2) errors get sent to stderr when called from the command line
3) various documentation problems
4) fixed display bug discovered by Alex van den Bogaerdt
   the graph display was off by one step
5) fixed infinite loop which occurs when trying to paint a grid line
   at 2:00 am on a day when day light saving time gets activated ... (now
   guess what happens today :-)

* From Blair
0) Fewer spelling errors for CHANGES file ... thanks !
1) some additional memory cleanup so that rrdfiles do not contain
   random date in unused areas ...

0.99.20
* From Oleg Cherevko <olwi@icyb.kiev.ua>
0) Allowed uppercase letters in DEF and CDEF names for rrd_graph (note
   that you now can produce conflicts when using strings which match
   the name of an rpn function call
1) make at style time code understand long weekday names
2) corrected documentation about the length of ds names (it 19 chars and not 9)

* From Tobi
0) Switched to calloc in rrd_create, to prevent from arbitrary strings
   showing up in the rrdfiles
1) Made the two graphs produced by the perl-shared example code match by
   specifying the end time in addition to the start time ... There is still
   a potential difference between the gifs, because there seem to be some
   random entries in the color map, which do not define colors for any pixels
   actually in the graph. If you convert the graphs to PNM format they are
   completely the same

0.99.19
* From Tobi
0) Fixed example in rrd_create man-page
1) Fixed rpn parser to properly track through the expressions ...
2) Fixed stacking code so that it is possible to stack the same variable several times.
3) Added --base switch to rrd_graph which allows to switch the scaling base to 1024
   this takes care of situations where you want to graph memory instead of
   traffic

0.99.18
* From Oleg Cherevko <olwi@icyb.kiev.ua>
0) Removed vsnprintf from parsetime for portability reasons ...

0.99.17 
* From Tobi
0) added the missing files from Olegs at patch ... now the thing
   compiles again ...
1) massively enhanced the number of functions accepted in the
   CDEF arguments of rrd_graph. Things like if, less than, larger or equal,
   is unknown and more are available. Check the rrd_graph documentation for
   more info on this ...

0.99.16
* From Oleg Cherevko <olwi@icyb.kiev.ua>
0) The rrdtool function HandleInputLine never checks the return values of the
   rrd_fetch() and rrd_graph() calls.
1) Added AT-STYLE TIME SPECIFICATION. This allows to specify the time
   using a more user-friendly format than all just seconds since 1970.
   See the rrdfetch manpage for a long description of these features ...
0.99.15
* From Blair
1) Uses CFLAGS to compile RRDs.o with the same optimization
   flags.
2) Remove some extra ;'s
3) Spelling fixes.

0.99.14 
* From Tobi
0) rrd_graph now returns the gif image size. The commandline version just
   prints XSIZExYSIZE after being called and the perl version returns
   the size as two additional return values from the graph call.

   PLEASE NOTE THIS mAY BREAK SCRIPTS ANALYZING THE DATA RETURNED BY
   RRD_GRAPH ...

1) added perl-piped/  perl-shared/ to make all building

0.99.13
* From: Oleg Cherevko <olwi@icyb.kiev.ua>
0) fixed typo in the prt_fctn code of rrd_graph ..

0.99.12
* From Tobi:
0) New graph function COMMENT added. It prints simple text into the legend.
   No evaluation occurs
1) New graph legend feature added. The legend area formating can now be
   influenced by adding special escaped characters to the end of some
   of the legend entries. Whenever such an entry is processed, the placement
   of the pending legend items is done as defined the special character.
   See NOTE 2 at the end of the rrdgraph manual for details.

0.99.11
* From Tobi:
0) added --template switch to rrd_update. This allows to specify which
   datasources you intend to update. It provides some protection against
   updating the wrong datasources.

0.99.10
* From Tobi:
0) added missing : in optarg template of rrd_create

0.99.9
* From Tobi:
1) removed rrd_format2.h this was not used ...
2) renamed diff to rrd_diff and added ranlib step to library
   creation ... this is all for the benefit of FreeBSD
* From: Andrew Turner <turner@mint.net>
1) added new consolidation functions LAST and TOTAL

0.99.8 Thu Feb 4 1999
* From Tobi:
1) Removed compiler waring from rrd_graph.c (blair, I went to default all
   the way ...)
2) revisited configure.in ... made gcc option test work properly

0.99.7 Mon Feb 1 1999
* From Tobi
1) removed kill command from the end of RRDp.pm
2) fixed rrdtune ... ds selector did not work ...
3) removed warnings from diff.c
4) fixed default -start time when creating new rrds ... 
   this was 1.1.1970 instead of NOW ...
 
0.99.6 Fri Jan 29 1999
* From Tobi 
1) more ... don't compare with nan fixes ... this time in rrd_update
2) on request from Russ Wright made VRULE and HRULE being painted on top 
   of the grid 

* From Blair
1) real decimal version numbers for perl modules
2) better argument checking for rrd_update

0.99.5 was only available to 30 minutes ...

0.99.4 Thu Jan 28 1999
* From Tobi
1) Happy BSD patch ... it seems that RRDTOOL will work on BSD once
   there are no compares with NAN ... I have altered one if in rrd_create.
   lets see if this helps

0.99.3 Wed Jan 27 1999
* From Tobi
1) ran with electric fence ... found bugs in RPN processing, and in data
  fetch (related to DS naming)
2) removed superfluous im_free calls which were in fact causing segfaults by
  trying to release memory which was already released ...

0.99.2 Tue Jan 26 1999
* From Tobi trying to find the cores reported by Jesper Skriver
  changed rrd_create to use DNAN instead of unknown. Maybe we have
  a problem with the representation of UNKNOWN as NAN on FreeBSD ...
  We should not get floating point exceptions because of this or should we ?
  
* From Blair
  1) Spelling mistakes in the pod files.
  2) Updated 0.99.0 to 0.99.1 in the RRDs.pm and RRDp.pm files.
  3) Updated the rrdcreate.pod file to accurately reflect the
     correct options to rrd_create().
  4) Allow uppercase characters in the ds-name in rrd_create.

0.99.1 Fri Jan 22 1999

* Fixed notice in gifs
* Fixed color selection for base elements 

Thu Jan 14 1999
*****************************************************************************
* BIG CHANGE ... The new format is in place .....
* drop your old files ... everything is new now .....
*****************************************************************************

* Tobi ... user visible changes are that the data sources do have
  names now and that the xxf can be set individually for each RRA
  this means that some command line options had to be changed ...
  Basically everything which deals with DS selection.

*****************************************************************************
*****************************************************************************
*****************************************************************************

Mon Jan 11 1999
* Tobi
1) Fixed data_reduce again ... blair pointed out a core dump condition in the
   perl module ... 

Sat Jan 9 1999
* Tobi
1) Fixed rounding error in rrd_graph/ytr ... it lead to zero values
   not being identified correctly ...
2) applied Philippe fixes for better compilation of the Perl module under NT
3) new version number for perl module
4) Fixed AVERAGE calculator in print_calc ... thanks to Sasha Mikheev for the hint
5) Fixed bug in reduce_data function which killed CDEF under certain circumstances because start
   was not adjusted according to the new step size ...

Wed Dec 16 1998
* Tobi
1) when using the rigid option and the data would be outside the graph, the
   line gets drawn 2 pixels out side graph ... this is not tested :-) 
   rrd_graph.c ~ line 246 around line 1900 I removed all the sanity checks for
   the graphics to be inside the valid area ...

* jeff alan
I'd like to be able to turn off interlaced gifs, since I think they
look messy when they are coming in. I added an option called
"interlaced" (abbrev: "i") which will set the graph to interlaced. The
graphs now default to progressive.

*phillipe
moved #include "../src/rrd_tool.h" in RRD.xs to etern"C" section
to make it work on NT ...


Sat Nov 21 1998
* Blair Zajac <bzajac@geostaff.com>
1) Update the rrd_update function to be able to take multiple
   timestamp/DATA pairs.  When there's a lot of data to dump
   into an RRD, this will save a lot of time.  In one test I
   ran updating 1000 data points into a single RRD trimmed
   40% off the execution time.  This reduction is caused by
   two factors: rrd_update gets called only once and the RRD
   gets opened and closed only once.  The new input to
   rrd_update still takes the old format, but allows trailing
   time and DATA: pairs:

   rrdtool update demo.rrd [--time|-t timestamp] DATA:...
      timestamp DATA:... timestamp DATA:... ....

   This change did a lot of work to rrd_update.c.  To see all
   the changes I would do a diff -ruWb between the previous
   and this version.  This will ignore lines that have had
   their indentation changed due to the addition of a while
   loop around the whole update structure.

   The only problem I see is that in the loop where the
   timestamp/DATA pairs are processed there is a single
   fwrite call to write some data to the RRD.  If some of
   these succeed and then one fails, should the remaining
   data that needs to be written at the end of rrd_update
   still be written to the file or should we just exit
   and leave the RRD alone.  If the first fwrite succeeds and
   everything else fails, will the RRD still be ok for
   future updates.

   I've run some comparisons between the RRDs generated by
   the old rrd_update and this rrd_update and I've found only
   a few instances where the RRDs differ.  The output of
   rrdtool dump did not differ however and the resulting GIFs
   looked identical.  I would recommend that we be careful
   with the code for a little bit.  I would greatly appreciate
   it if somebody else could take a look at it.  I do think
   it's important, as that 40% savings in time is very important
   for how I'm using RRD.

2) Update the POD for rrd_update.

3) Some spelling corrections.

4) Update test.pl to output demo1 and demo2 RRDs and GIFs.
   Demo1 is created using single timestamp/DATA pair calls
   to rrd_update while demo2 is updated using one single
   call.  This allows us to compare the output RRDs and GIFs.

5) To allow the comparison of RRDs created and updated using
   slightly different code, I found a problem where the filler
   bytes the compiler uses to align doubles and other types
   in the structs in rrd_format.h would be filled with random
   data.  This would cause simple cmp compares to fail on
   newly created RRDs.  Now rrd_create does a memset to 0 of
   all newly malloc memory to set these filler bytes to a
   consistent value.

Wed Nov 18 1998
* Tobi
1) rrdtool graph will now consolidate data previous before plotting it, if
   the resolution of the data is higher than the resolution of the plotting
   area.

Tue Nov 17 1998
* Tobi
1) allow negative --start time in create. This means go back the specified 
   amount of time from NOW ... this is the same as it is supported with other
   modules ... 
2) make sure we do draw outside the appointed area. With the --rigid
   option this could be accomplished and would make gdlib work so hard that
   rrdtool appears to be dead ..

Sat Nov 14 1998
* Blair Zajac <bzajac@geostaff.com>
1) 1000 spelling fixes (thanks!)
2) a new Datasource type called DERIVE. It calculates the derivative between
   the current and previous value of the datasource.
   This lets you plot the rate of increase or decrease of a counter
   value, such as the number of people in a room.
3) some contents fixes for the pod docs ...
4) more tests for the binary perl module

Fri Nov 6 1998
* Blair Zajac <bzajac@geostaff.com>
  more spelling and other fixes ... especially a few remaining exits were
  replaced by returns ...

* tobi
1) fixed the parsing of : in string arguments of rrd_graph. An output string
   can now contain a : if it is escaped with \ .. a \ has to be escaped as
   well ... this makes the changes from the 24-oct really work ...
2) in connection with 1) there were also problems with arguments where the
   string was blank ... 

Tue Nov 3 1998
* Blair Zajac <bzajac@geostaff.com>
1) added a missing fclose to rrd_open ...

Mon Nov 2 1998
* Tobias Oetiker
1) configurable color for arrow in Graph
2) added new format definition header to distribution .. this is bound to
   be integrated ... once I get some time to do it ... (any takers?)

Sat Oct 31 1998
* Win32 patches from Philipp Simonet <philipp.simonet@swisscom.ch>
1) New Visual Studio Config files for building WIN32 binaries
2) Locking enabled under WIN32

* Tobias Oetiker
1) fixed scaling again ... when there is no data, the graph will show 0..1
   and not -1..1
2) Added a little arrow indicating the positive x axis direction
3) paint x and y axis before the graph, so that it does not hide it

Sat Oct 24 1998
* Perl and other patches from Blair Zajac <bzajac@geostaff.com>
1) Have rrd_last return a time_t instead of an int, so the real
   time from the RRD file is returned.
2) Fixed an error where rrd.live_head->last_up was returned from
   rrd_last after rrd was freed.
3) Make perl/Makefile.PL a little neater.
4) Add a realclean option to perl/Makefile.PL.
5) Small formatting change to RRD.xs.
6) Added a little RRD::last testing code to test.pl and added some
   more ok's for now and future use for more tests.  Changed the code
   to run under use strict.
* Tobi O.
1) removed machine option from rrd_last ... it will now only return
   the unix time_t when the rrd was last updated ... 
2) allo ':' in graph arguments, if the : is escaped with a \. In perl
   strings you have to enter two \ because perl interprets them 
   as well: eg  "Max\\: %2f"

Fri Oct 23 1998
* Makefile Patches from Blair Zajac <bzajac@geostaff.com>
1) Allow the user to do something like make CFLAGS=-O3 and not loose
   the other important flags, such as -DHAVE_CONFIG_H, etc.
2) Move all OS and compiler dependent flags into CFLAGS_EXTRA.  This
   changes Makefile.in's and configure.in.
3) Add realclean targets to all Makefiles.
4) Have realclean clean up many more things to return the package to
   a clean state.

Mon Oct 19 1998
* Fixed rrdcreate.pod ... the example was out of date ...

Sun Oct 18 1998
* From Jeff: rrd_open will open readonly when ever possible. This
  allows to graph without write permission ... 
* From Jeff Allan: Removed all remaining  \n from the error
  messages and fixed some spelling
* From Jeff: Improved handling of cases where no data is
  found
* rrd_graph: if minval == maxval pumb max up and pull min down 1% 
  ... by tobi as suggested by allan.

Mon Oct 12 1998
* This time the diff.c fix for negative numbers
  and counter wrap detector code are really in ... tobi

Thu Oct 08 1998
* Added file locking code to rrd_update ... written by Russ Wright.
  The locking code uses fcntl so I guess it will not work on NT ...
  any takers ?

* Fixed longstanding bug in diff.c ... now it works even for 0-10
  Tobi.

* made rrd_graph work even when there is no data to plot ... Russ Wright

* added new last function provided by Russ Wright. This func allows
  to query an RRD for it's last modification date ... 

* added counter overflow patch as suggested by Andres Kroonmaa <andre@ml.ee>
  to rrd_update. This will only work for 32 bit and 64 bit counters.
  Are there any others in SNMP land ?

Thu Aug 13 1998
* raised stacksize in rrd_graphs rpn machine to 30 (otmar)
* fixed label placement in graph for precisions != 0 (otmar) 

Sat June 13 1998
* added ms devstudio stuff from philippe

Mon June 8 1998
* WIN32 Portability fixes from Philippe

Sun June 7 1998
* fixed coredumps with rrd_error ... I was freeing free memory.
  somehow I had assumed, that free(x) would set x to NULL ...
  this is not true :-)

Wed June 3 1998
* fix for PRINT and GPRINT calc in rrd_graph ...
* fix for STACK function command line parsing ... It must follow
  directly after a LINE*, AREA or STACK function

Tue June 2 1998
* lots of fixes from Jost Krieger integrated
* increased MAX_LENGTH to 10000 in rrd_tool as suggested by simon ...

Tue May 26 1998
* did some instruction reordering in rrd_graph.c to make
  a segfault go away which was discovered by David Vickers

Sun May 24 1998
* cleanup session. In order to improve rrdtools perl integration
  I have started moving all the printf code out of the individual
  functions. Now they all return their findings (except rrd_dump)
  to rrd_tool.c which then prints them.

* the perlmodule got extended to use these new possibilities. The
  RRD::fetch and RRD::graph return the things they used to print
  directly to perl. See the test.pl for an example. There is still
  no documentation though for the perl module
  
Tue May 12 1998
* Patch to rrdcreate.pod from Amos Shapira <amos@gezernet.co.il>

Thu May 7 1998
* debugged memory allocation and access errors in CDEF. Now
  things should work as advertised ... Thanks to
  David Vickers <David.Vickers@citec.com.au> for finding the
  bug ...

Fri May 1 1998
* finally added the --logarithm switch ... or rather the logic behind it.
  you can now draw graphs with logarithmic scale in the data axis ... 

* fixed some minor drawing bugs ...

Thu Apr 30 1998
* hp portability patch from "Crawley, Tom" <Tom.Crawley@hi.riotinto.com.au>
  you should now be able to compile with gcc on hp as well ...

Fri Apr 18 1998

* Patch from Otmar Lendl (O.Lendl@Austria.EU.net) to
  fix perl coredump ... 

* Split the top makefile into Makefile.dist and Makefile
  Makefile does not contain any gnu only things anymore

* Renamed doc Makefile to GNUmakefile because it contains
  several GNU only constructs. Lets be straight in this respect
  and stop pretending it was a Makefile. 

Thu Apr 9 1998 
**** INCOMPATIBLE CHANGES ****

* altered PRINT and GPRINT functions ... we really need to give
  them a consolidation function ... [G]PRINT:<vname>:<cf>:<format>
  otherwise things are not clear for calculation. You can now
  define the operation for PRINT and GPRINT independent from the
  data source consolidation method. This allows to calc things
  like average maximum, or maximal 5 minute average ... or what ever
  you choose ...

* options of graph, create and update have changed again. The old style
  LINE1:..., DS:..., DATA:... arguments are back. I found that this
  is more logical than doing everything with options. This will be
  especially helpful for the perl interface

* Completed CDEF/DEF implementation. Now you can create new
  data sources on the fly by doing some rpn math. Check rrd_demo.pl for
  an example of CDEF rpn math.

* updated rrdgraph, rrdupdate and rrdcreate  man-pages to reflect new
  command-line options

* updated rrd_demo.pl to work with new command-line options
  log2rrd.pl and graphrrd.pl are *NOT* updated yet ...

* updated perl test.pl to work with new options ... unfortunately
  test.pl crashes after generating about 3 graphs ... I guess there is
  a memory leak, but debugging the core of an shared library perl
  is quite a problem ... no solution yet ... When I link perl statically
  to the rrd stuff things work without core ... 

Wed Apr 8 1998 -- Wrolf / Russ
* new scripts log2rrd.pl and graphrrd.pl 

Mon Apr 6 1998
* compete RPN implementation ... things compile ... too tired to
  test, will do later

Sun Apr 5 1998
* added -fpic for gcc and -z for hp compilers to generate
  position independent code which helps when doing shared
  libraries ...
  
Thu Apr 2 1998 -- Wrolf
* Makefile.in fixes

Wed Apr 1 1998
* Updated docs for rrd graph ... will take the graph functions out of long
  opts ... as they are not really options ... 

Tue Mar 31 1998 
* small Makfile.in fixes ...

Tue Mar 31 1998 -- Wrolf
* spelling fixes 

Mon Mar 30 1998
* big source split ... its now all neatly separated
  in subdirectories.

* gd1.2 is now included in the release ...

Mon Mar 30 1998 -- Philippe.Simonet@swisscom.com
* rrd_open.c use rb+ instead of r+ for fopen
* ntdemo.pl created to the test rrdtool.exe under NT

Sun Mar 29 1998
* INCOMPATIBLE CHANGE remaining functions converted to getopt_long
* started loadable perl module ... its all in the RRD directory

Thu Mar 26 1998 -- jeff.allen@acm.org
* added some files to make things work under DevStudio
  not tested yet ...

Tue Mar 24 1998
* started changing calling sequence rrd_graph.c done
* updated to much newer getopt ... maybe it has a few potential 
  bugs fixed
* updated rrdgraph.pod for new calling sequence

Mon Mar 23 1998 
* improved labeling for horizontal grid

Sat Mar 21 1998 -- jeff.allen@acm.org
* hpux portability improved
* sunos port

Wed Mar 18 1998
* added log2rrd.pl to configure.in and Makefile.in
* about 1'000 spelling fixes from Wrolf applied 
* Fixed help screen in rrd_tune (LINE1 ...)
* Fixed error in GPRINT and PRINT function of rrd_graph.c (Philippe)
* added log2rrd.pl by Russ and Wrolf

Sun Mar 15 1998
* Updated Makefile to produce *.pl.in files from *.pl files on tar and dist
* Updated Makefile to generate *.html and *.man files from *.pl and *.pm 
* total rewrite of horizontal grid lines I guess now things look more the
  way they should ...
* INCOMPATIBLE CHANGE the -y option is now called -v and -y is now
  equivalent to x in the respect, that it allows to configure the horizontal
  -grid. (log grid will come soon)
* implemented title function -t (it was already there but not active)
* made RRD.pm complain if it is not used correctly
* updated documentation for RRD.pm
* INCOMPATIBLE CHANGE ... there are now 3 types of LINE -- LINE1 LINE2 and
LINE3
  they are increasing in size.
* if there are only PRINT commands in a graph command line, no GIF will be
  created ...

Fri Mar 13 1998
* fixed some potentially ambiguous else statements (gcc-2.8.1 objected)
* fixed error in STACK option
* added documentation for log scale
* added documentation for y-axis spacing config

Thu Mar 12 1998
* added rule to configure so that it looks in
  /usr/local/include/gd for include files
* added -d option to rrd_tune.c to allow to change
  the DST of a DS ... (special request from Alan)

Thu Mar 12 1998 -- jeff.allen@acm.org
* cleaned up rrd_tool.h to make better use of configure
* tried to separate out Win32 stuff better
* made calls to strftime fail if it doesn't exist (probably
  need to ship a substitute, unless this is more portable than it sounds)
* created config.h.nt, which will probably be useful for porting
  to NT. We'll see.

Wed Mar 11 1998
* oops nonblocking was NOT in last release ... sorry ... here it is ..
* started to create RRD.pm 
* adapted rrd_demo.pl to work with it ... 

Mon Mar 9 1998
* better values for xlab array ...
* fixed error with rrd_fetch 
  (it was not fetching the right data if the pointer wrapped straight from
   the beginning)
* check VRULE and HRULE placement
* changed linking order -lgd and then -lm as floor from gd is in m
* made rrd_demo use nonblocking read to get data from pipe

Mon Mar 9 1988 -- Wrolf (how about using diff -u )
* lots of spelling fixes
* some linting fixes
* better Makefile portability

Sun Mar 8 1998
* added rrd_tune.c and rrdtune.pod ... read the docs ... 
* compiled rrdtool and gdlib with -pg in order to run it
  under gprof ... the results were, that most time is spent
  compressing the gif file. As the profiling is only on a function level, I
  was not able to identify hot spots within the code ...
* added getrusage and gettimeofday  call to rrd_tool.c in order to get
  better grip on time spent on each function ... the results will be printed
  after the OK prompt. Adapted rrd_demo.pl accordingly ... 

Sun Mar 8 1998 -- jeff.allen@acm.org
* Fix to first line of rrd_demo.pl. Autoconf docs say that
  Sequent's unix needs the space. Whatever...
* cosmetic changes to rrd_demo.pl.in
* changed getopt.c to suck in rrdtool.h, so that it includes
  string.h. The comments in getopt.c seem to imply that
  this can break things, but it clears up a warning for me.
  What do others think?

Sat Mar 7 1998 
* tried some cool graphs ... the current one produced by rrd_demo.pl
  looks really freaky ... not that anyone would want to use
  such a graph, (I guess) but it can be done ... 
* finally rrd_graph does work as described in the manpage
* many cosmetic changes to the graph
* use Open2 and select in rrd_demo.pl
* added OK response to rrd tool
* integrated simons opt patch ...

Fri Mar 6 1998	-- jeff.allen@acm.org
* docs in configure.in
* HP cc and GCC flags discovery (not tested on HP -- Wrolf,
  please give it a try)
* improvements to makefile
* one-line changes to rrd_graph.c and getopt.c to keep Purify and
  Gcc happy.

Thu Mar 5 1998
* some more HPUX portability stuff by Wrolf
* rrd_tool.h cleanup
* added project files for MS VC++ by Philippe
* aesthetic tuning on the graph ... 
* text labels below the graph area
* VRULE and HRULE implemented
* Debugged argument parsing ... no more segfaults when you give the wrong
  number of arguments.

Wed Mar 4 1998
* now rrdgraph is working ... not all methods are implemented yet, but it
  does produce graphs ...  see rrd_demo.pl 

Tue Mar 3 1998
* fixed some further minor problems .... bug fixing goes ahead, but not
  finished yet ...  CODE DOES NOT WORK

* found the 'memory problem' there is a difference between
  malloc(a-b * sizeof(c)) when I want to do malloc((a-b) * sizeof(c)) 
  I love braces ... :-)

Tue Mar 3 1998 -- jeff.allen@acm.org
* Added configure.in and friends, to simplify configuration.

Sun Mar 1 1998
* started to make rrd_graph work as advertised. Programming went well,
  unfortunately debugging is not yet done and not all the functionality
  is implemented. But I guess if it did not crash, the AREA, LINE and
  STACK functions should be OK ... I guess there is some memory problem.
* enhanced Makefile to be able to do the man and html versions

Thu Feb 26 1998
* fixed pipe mode commandline parsing
* made rrd_demo.pl work again ... no graphics at the moment
* altered commandline options for update ... negative numbers were a problem
* Integrated some patches from Wrolf 
* filename and pod patches from Jeff added
* changed Makefile structure to make it ready for autoconf/configure ... (jeff)
* added rrd_getopt for good now it will work the same on all systems.
  Maybe we should go to gnu_getopt once jeff integrates configure ?
* made rrd_update work as advertised in the manpage

Wed Feb 25 1998
* some 'make the compiler happy' patches from Jeff
* CFLAGS for HP cc from Wrolf
* included source for getopt in case your system does not have it 
  edit Makefile ...

Sun Feb 22 1998
* improved/spell-checked pod documentation (Russ)
* rrd_fetch and rrd_dump work as advertised in the docs

Sat Feb 21 1998
* stared bringing code in line with documentation. rrd_create.c done

Sun Feb 15 1998
* created full documentation for rrdtool in pod format

Sun Feb 1 1998
* fixed seek bug in rrd_update
* rewrote rrd_tool argument scanner now it does quoting as it should be.
* rrd_graph ... scaling bug fixes
* horizontal grid fixed
* added copyright notice
* vertical grpd fixed
* wrote demo app in perl
* updated TODO and README

Thu Jan 22 1998
* Russ has debugged my rrd_graph code. 
* Fixed one remaining malloc error
* improved visual appearance of the graph
* updated TODO list

Sun Jan 18 1998
* added Dan and Russ rrd_tool.c ... rrd_main.c is gone
* current setup assumes gd1.2 to exist in the compile directory
* things compile again ... 
* rrd_graph is done, but not yet debugged ... so while it compiles,
  it will not do what it should (I guess )
* this release is for dan dunn to integrate the new rrd_main.c
* created TODO and CONTRIBUTORS files

Thu Jan 08 1998

* mrtg DOES NOT COMPILE ... you can leave out rrd_graph and the
  relevant parts of rrd_main to get it to compile ... 

* releasing the code because there were some changes in other modules as
  well 
* rrd_graph is about 50 percent done ... remaining twist clever
  horizontal grid ... 

Sat Nov 29 1997

* debugged rrd_fetch 

Fri Nov 28 1997

* rrd_update a ; after a if prevented CF MAX from working

* rrd_dump fied time in output

* rrd_perf fixed mrhb in demo db

* rrd_fetch created. This function will get the data
  for the graphs

Thu Nov 27 1997

* rrd_update debugged ... I guess (!) it works now correctly

  note: we always assume that a input value represents the average
        from the last to the current run

* rrd_fupd created. This is like rrd_update, but it accepts input
  from a file or stdin ... (this saves us from restarting
  rrd_update for each file)

* rrd_perf.pl created. It helps to assess the performance you
  can achieve with an rrd on your system.


Sun Nov 23 1997

* rrd_update created ... does not yet work ... but it should !
* some problem s with the format removed in the process ...

  
