We are in the process of updating the archive from Eprints 2.2.1 to 2.3.3. The new version has some new options, one of which is the possibility for splitting the submission proces into more stages, see

• Submission Stages

Below is described the major changes from version 2.2.1 to the new 2.3. versions. In addition every new version fixes minor bugs and typos. Note that not all the new options can be easily implemented in the Organic Eprints archive because they involve changes to the database fields structure.

EPRINTS 2.3.3 2004–03–04

 - Added explain_sql_tables script which explains what all 
    the SQL tables are used for. Handy for hacking and debugging.
 - Fixes to generate_views handling of subjects fields.
 - Fixes for Apache 2 compatibility
 - The name of a language can now be specified in an language specific 
    phrase file. Added a new method to session: render_language_name
    which renders the given name in the right language. 
 - The string used to join names (and other field types) can now be
    varied for the first or last item. So you can join names with “, “
    but for the last pair use “ and “.

EPRINTS 2.3.2 2004–02–25

  - Fixes for MySQL 4 compatibility.
  - Fixes some problems with generate_views handling of subjects.
  - Fixes for Apache 2 compatibility.

EPRINTS 2.3.1 2004–02–05

 - “last indexed” information to search results page so that users won’t
    be (as) confused if a search does not contain an item they just 
    added.
 - Browse by ‘name’ fields now works. So you can have a static “browse”
    page for each unique author name eg. Fenderson,_Marvin.html - this
    is not perfect as he may also by Fenderson,_M.html etc. Or you 
    might have to Marvin Fendersons. Browse by author ID is still a
    safer bet in the long term.
 - Hooks so thta input fields can vary based on other factors. This 
    should allow input fields which can (for example) limit the options
    in one field based on values in another.
 - Fixes for Perl 5.8 compatibility.

EPRINTS 2.3.0 2004–01–12

 - Date fields may now be configured so that only the year part is
    required, with month and day optional. They may be treated as
    if they were “year” fields for browsing and rendering citations.
 - Completely new default fields, help and citations. Based on the
    work of the TARDIS project.
 - The registration process has been cleaned up a bit. Now change
    password and register for new account are two seperate forms.
    Also, fields from the user dataset can be added to the register
    page (eg. Name), to save doing it later.
 - If a user record is not valid the users home page now automatically
    redirects to the edit my details page.
 - Changed layout of the users home page. Cleaner layout.
 - Metadata “types” code completely rewritten so as to be object-
    oriented. Eg. Url extends Text. This should allow sites to add
    their own custom types with unique rendering, input, storage
    and search properties.
 - Filters. As described for searches filters allow you to limit
    the results of searches to only those matching a hard-wired 
    setting. Filters can also be applied to the OAI interface and
    latest_tool and will be applicable to views in a subsequent 
    version.
 - Citations are now full of <span> tags indicating fields and
    also peoples names. Handy for style-sheeting.
 - Apache 2 support.
 - OAI Config now roughly follows Andy Powell’s recomendations. nb.
    This only affects new archives.
 - Multilang archives can now understand the browsers language 
    preference.
 - Views may now have sub-pages eg. /year/2000/article.html 
    (browse by year then type). 
 - Views may now have sub-headings. eg. Browse by year, but results
    page is divided into <h3>Article</h3>…..<h3>Book</h3>…etc.
 - New CGI Tool: latest_tool. This lists the last “n” additions to
    the archive. It can also have more than one “mode”. Eg. Another
    mode might be to output the last “n” items which had full
    texts attached. It can also be made to output results in
    RSS format.
 - New CGI Tool: counter. This just lists the size of each dataset
    in a simple, machine-readable way. This is to allow eprints.org
    to monitor growth of archives in a low-impact way.
 - Option to work with Apache 2. Apache 2.0 support relies on
    3rd party libraries which are in active development. 
 - Tested with perl 5.8 on Redhat 9.0. Some bugs fixed with unicode
    support problems.
 - Mime Tex? support.

Submission & Data Input

 - Option to make the eprint type selection a list with radio buttons
    and long descriptions, rather than just an HTML option element.
    (submission_long_types). This is the default for new archives.
 - Paged metadata input. Rather than overwhelm the user with a huge
    form, the eprint metadata input may now be split into several
    small stages.
 - New input_style for “set” and “datatype” fields: “long”. This
    renders them as a deliniated list with radio buttons and long
    descriptions taken from the phrase file. (This is the option 
    which “submission_long_types” turns on for the system field
    “type”.
 - The input method for fields can now be completely replaced with
    alternate code. A much nicer input method for “subjects” is 
    supplied.
 - Changed the layout of the submission form. Now has control buttons
    at the top and bottom with one being “save for later” (rather 
    than the confusing “cancel”). The file upload section is now 
    much quicker and cleaner. It does not even ask what the
    “primary file” is unless there are two or more files in the 
    document.
 - “Help” for metadata input fields can now have HTML elements in
     including links or javascript pop-ups. Different help may be
     supplied for different types, eg. The help for title-of-article
     can be different from title-of-book.
 - Improvement to input-form-render for metafields. The layout is
    now much more compact. Multiple fields now have arrow buttons
    to allow them to be easily re-ordered.
 - Improvements to the V Lit? transcluding interface.

Searching & Indexing

 - Search Field help may now contain HTML.
 - Indexing and search code rewritten. Now indexing is done by
    a daemon which runs in the background. This means that you don’t
    have to wait for the indexes to keep updating while editing a 
    live record.
 - Web-interface searches redesigned. No longer “simple” and 
    “advanced”, but as many different search configurations as you 
    need. Searches can have custom citation style, items-per-page,
    title, help and can even have “filters”. Filters allow you to,
    for example, limit a search to papers where status=“published”
    without that being an option in the form.
 - Full text search tool. This uses the bundled Convert::Plain Text?
    module. That just provides a wrapper to various UNIX tools which
    extract the words.
 - New name search syntax. “Smith, R” will match “Smith, Robert” 
    and a search for “Smith, Robert” will match “Smith, R”.
 - Searching fields of type “date” now possible. Allows searching
    of just years or exact dates.
 - Staff searches are now just like the new user searches. By default
    there are searches to search the user, buffer, inbox, archive
    and deletion datasets, but more can be added.