Changes in MRS version 6.0.5

- Changed e-mail address of maintainer.
- Added base tag to error page and fixed handling of GET on webservices URL.
- Bugfix issue #1: New jobs are removed from queue to soon resulting in UNKNOWN
  status.

Changes in MRS version 6.0.4

- Fixes for building on CentOS
- Fixed some license issues

Changes in MRS version 6.0.3

- Bug fix in Blast web service, incorrect handling of optional Parameters
  block resulted in crashes.
- Restart of server should now be more robust using timed_join for server
  thread.
- Max run time for external tools (clustalo e.g.) can be set in config file.
- Better handling of time out for clustalo.

Changes in MRS version 6.0.2

- Several web service changes to make implementation of MRS::Client easier:
	- removed GetIndices, data is now in GetDatankInfo
	- aliases are now accepted in all calls
- Renamed m6 to mrs. Which is a bit more consistent name.
- Various changes made to installation due to creating a debian package.
- Changed some code to work around weird gcc 4.7 crashes.

Changes in MRS version 6

- Single file databanks have been replaced with one directory per databank.
- There's only one executable left: m6. This executable takes as the first
  parameter a command name and then options for the specific command.
- Configuration is done in just one, xml file. This can be edited using the
  admin interface in the web application.
- Only one server instance and only one IP port used.
- Makefile (mrs-update) has been replaced by a scheduler inside the web
  application.
- Parser files have been simplified, they now only contain all that's needed
  for parsing and nothing more. The configuration previously contained in the
  parsers has moved to the single configuration file. There's also a generic
  parser that can be used to index 'one document per file' databanks.
- The web app no longer formats entries, this is now done by formatters
  written in JavaScript and handled by the browser. Unfortunately this means
  that Webkit based browsers can no longer display PubMed Central records,
  neither can Opera. This has to do with a bug(?) in the handling of XML 
  Stylesheets in these browsers.
- New link implementation. The parser takes care of recording links to
  external documents and the server takes care of the rest. No need to
  explicitly define links between databanks anymore.
- New REST interface, the old one was dropped.
  To retrieve the fasta for the 104k_thepa record from SwissProt, you call:
  
  	http://mrs.cmbi.ru.nl/m6/rest/entry/sprot/104k_thepa?format=fasta

  of course, you can leave out the format parameter, in which case you
  get the plain document.
  
  To search for crambin in swissprot, you would call:
  
  	http://mrs.cmbi.ru.nl/m6/rest/find/sprot/crambin
  
  Here, you can optionally specify the parameters 'offset' and 'count', as in:
  
  	http://mrs.cmbi.ru.nl/m6/rest/find/sprot/crambin?offset=1&count=2

- Joined databanks have been dropped. This was a nice way to overcome the
  fact that MRS files cannot be updated with new or modified documents.
  However, the performance penalty was large. The alternative is to use
  aliases that can specify a group of databanks. E.g. uniprot is an alias
  for both sprot and trembl.
- The tokenizer has changed slightly to optimize recall.
- A new way to browse index entries. Go to the information page for a databank
  and click on the index names. This will open a page where you easily select
  index entries and the documents that contain them.
- NCBI blast support has been dropped, the MRS blast code has been optimized
  and now works on FastA files directly. The blast cache is now stored on
  disk and the index is managed by a SQLite databank. Results are removed
  once the databank they point to is updated.
