# ScanErrLog (c) 2000-2001 Jerome Alet & Free Software Foundation
#
# $Id: NEWS,v 1.31 2001/04/23 15:21:32 jerome Exp $
#

ScanErrLog NEWS:
================

	2.00:	- Version changed to reflect the need of the jaxml-2.22 (at least)
		  module.
		- All references to the jahtml module were suppressed, because
		  all is now done with the jaxml module.
		- Now works fine with ReportLab 1.06

	1.9 :	- A missing import caused the program to crash under
		  at least Solaris.

	1.8 :	- A bug was corrected when using the --continue option
		  and reading the error messages from stdin: if you
		  relaunched ScanErrLog then an illegal seek error was
		  possible under some circumstances.

	1.7 :	- Optimisations in the reading of the error_log file.
		  Should be faster but consumes more memory.
		- The -c | --continue option was added. This should
		  be very useful to parse very big error_log files
		  multiple times.

	1.6 :	- Very long lines are now dealt with correctly
		  in PDF format (requires ReportLab 1.01):
		  too long messages are replaced by (...) in
		  red color.

	1.5 :	- Now requires the jaxml Python module,
		  you have to download it separately.

		- The jahtml Python module is not
		  distributed anymore with scanerrlog,
		  you have to download it separately.

		- Now uses Distutils as the installation
		  method, see README for details.

	1.4 :	- A nice report name is now proposed
		  automagically when used as a CGI script.

		- < and > characters are now automatically
		  replaced with their equivalent SGML entities
		  in HTML and XML reports as well as during the
		  creation of PDF reports.

		- a minor change in the XML report.

	1.3 :	- PDF support is completed, thanks to the
		  reportlab team for the splitting tables.
		  Very long messages may look ugly in a
		  PDF report, wait for next versions.

		- Lots of cosmetic changes (maybe bad ones)
		  in the program.

	1.3b3 : - Some minor changes in the documentation.

	1.3b2 : - PDF support was added. Because of some
		  limitations in the Reportlab library,
		  creating a document with a table which cover
		  several pages doesn't work. Hopefully all of
		  these limitations should be removed soon by
		  the Reportlab team (http://www.reportlab.com)
		  That's the reason why this version is numbered
		  as beta and not final: it works, but with
		  a not yet existing version of Reportlab :-(

		- you can choose the output format when using
		  ScanErrLog as a CGI script

	1.2 :	- the documentation on how to install ScanErrLog
		  was updated.

	1.1 :	- the XML format has changed a bit.

	1.0 :	- a CREDITS file was added

		- You can choose the output format with the
		  new -f | --format option. Allowed output
		  formats are: HTML (default), TEXT, and XML.
		  PDF support is planned. You can look at
		  SEL-Report.txt and SEL-Report.xml for sample
		  reports identical to SEL-Report.html

		- The ApacheErrorLog class defines a public
		  method per report type, e.g. :
			htmlreport()
			cgireport()
			xmlreport()
			textreport()

		- You can now exclude some messages from the final
		  report based on their severity, with the new
		  option -e | --exclude. This is easier than
		  doing it by rewriting the regular expression.
		  The --exclude option can be used from scanerrlog
		  when launched as a CGI script while the user
		  supplied regular expression cannot.

		- The ApacheErrorLog class constructor now accepts
		  another argument, exclude, which is a string
		  representing a slash separated list of messages
		  severity to exclude from the final report.

		- The HTML report background color was changed
		  from Pink to White.

		- Incorrect file permissions, for the sample
		  reports, were corrected.

		- Error messages are now stripped.

	0.9 :	- The report is now generated in HTML by default,
		  look at SEL-Report.html to see a sample report.

		- You can now use ScanErrLog as a CGI script, look
		  at ScanErrLog.html to see a sample form.

		- You can define your own title for the report with
		  the -t | --title option

		- You can suppress the HTML header and footer tags
		  with the new -w | --withoutheader option. This
		  may be useful if you want to directly include
		  ScanErrLog output in your own HTML document.

		- A new option -o | --outputfile was introduced to
		  allow people to save the report in whatever file
		  they want.

		- The -c | --cumulate option is not available anymore
		  since it is now the default behavior. Use the new
		  -n | --nocumulate option to restaure the old
		  default behavior.

		- The README file now includes some hints for the
		  installation.

	0.8 :	- An error in the statistics at the end of the README
		  file was corrected. There's no error in the statistics
		  really produced by ScanErrLog, but this particular one
		  was computed by hand, and I'm bad at maths ;-)

	0.7 :	- A test is made to detect if your Python distribution
		  or your standard C library supports the strptime()
		  function. If not then no date will be included in
		  the final report.

	0.6 :	- The --date and --pattern are now mutually exclusive
		  because a user defined pattern may eventually select
		  lines without any date in them.

		- The --date option was modified to display dates
		  between parenthesis just after the occurences count.
		  It's now easier to read.

		- Some statistics are now included in the final report.

		- Some code rewrites, it's probably easier to read and
		  should consume less memory (untested).

	0.5 :	- ScanErrLog now accepts a new option on the command
		  line:
		  -d | --date		  include in the final report
					  the date when each message
					  appeared for the last time.

	0.4 :	- ScanErrLog now accepts a new option on the command
		  line:
		  -l | --limit lim	  selects messages only if their
					  number of occurences equals or
					  exceeds lim.
					  lim's default value is 1, meaning
					  all messages are included in the
					  final report.

	0.3 :
		- ScanErrLog now accepts arguments on the command
		  line. They are:
		  -c | --cumulate	  to cumulate for all the files
					  passed on the command line.
		  -p | --pattern regexp	  to select the lines to parse
					  using this particular regexp.
		  -h | --help		  displays the help
		  -v | --version	  displays the version number

		- The default pattern to select lines was modified
		  because it was "not so common"

		- The __repr__ build should be faster

		- The ApacheErrorLog class now accepts a second
		  optional argument pattern in its __init__() method.

		- Reading the source's main body shows all the different
		  possibilities of this module/class.

	0.2 :
		- ScanErrLog now only keeps "normal" lines, and
		  displays at the end the number of skipped lines.

		  A normal line is something which looks like:
		  httpd: [date] [severity] [client] message: details

		- If you don't pass any argument on its command line,
		  nor redirect its standard input, ScanErrLog now
		  displays its documentation.

		- a NEWS file is now included (this file)

		- ScanErrLog is now managed with CVS

	0.1 :
		- First available version

Jerome Alet - <alet@unice.fr>
