EPrints 3.4.4

This page contains information about the provisional EPrints v3.4.4 tag and release on GitHub. A date for this release is planned for the end of June 2022.

Provisional Release Notes

• Zero codename: To be determined
• Publications flavour codename: To be determined

New Dependencies

None. Check earlier dependencies for EPrints 3.4.3 and before.

Changes Since 3.4.3

New Functionality

• Adds new script to about unit test results in JUnit format to better support use with Continuous Integration applications (e.g. Jenkins).
• Adds function $document->get_derived_versions to get derived documents, such as preview and thumbnail images.
• Allows custom list of fields for meta tags in summary pages.
• Allows parts of the subject tree to be expanded by default.
• Allows top of subjects based view to be different from field's top. (E.g. have a browse view for a sub-division of the organisation rather than the whole organisation).
• Adds show_timestamps_in_log config option to log.pl config files.
• Adds --reload flag to generate_static to also reload the repository configuration. This ensure changes to both static pages and templates are applied at the same time.
• Adds function to check whether a dataset with a particular ID actually exists.
• [https://github.com/eprints/eprints3.4/issues/201 Adds redo_hash option to epadmin to allow MD5 hash of files to be recalculated, if something when wrong with this originally.

Security and Privacy Improvements

• Ensures render_fileinfo always uses https URLs if enabled. So files can be downloaded securely.
• Allows EPrints::System::proc_exists to work when SELinux is enforcing.
• Ensures access settings (i.e. security field) of all derived versions of a document are updated when the document's access settings are updated.
• Ensures workflow cannot update a data object with anything other than a POST request.
• Adds monitoring for login attempts for better intrusion detection. Stored as hierarchical log files (e.g. 2022/04/13/) in login_attempts sub-directory of archive's var/ directory.

General Improvements

• Includes check of dark_document dataset before assuming no document in certain position on eprint is present.
• Allows bespoke function ultimate_doc_pos to be defined to work out position for new document, if eprint can use more than one document dataset (e.g. document and dark_document).
• Allows saved search alerts to go to additional recipients.
• Adds title phrase and warning to /cgi/set_lang if it were ever to be called without an immediate redirect.
• Allows CSS to be applied to specific subjects in a tree of Subject field.
• Send test email multiple recipients.
• Adds high resolution timestamps to database debug lines.
• Deal better with field type mismatches in SWORD POST requests.
• Allows epadmin reindex, recommit, redo_thumbnails and redo_mime_type to specify a range / multiple eprints
• Add timestamps to significant error messages.
• Allows distinctby search condition if fields if they are sub/super classes (i.e. types) of each other.
• Adds DOIs to OAI_UKETD_DC export plugin.
• Makes progress on improving Pod documentation in code files to update API wiki pages.
• Removes no longer implemented misc.pl configuration settings.
• Merges all RDF URI configuration files to tidy into a single file.
• Merges user_password_maxlength.pl into user_auth_limits.pl.
• Removes flowplayer as Flash is no longer supported.
• Removes defunct <t>ie6.css stylesheet file.
• Replaces old bespoke theme with leaner example theme.
• Ensures summary page EPrint::Box plugins have unique class for CSS.
• Provides a more Accessibility compliant version of limit_names_shown.pl.
• Extricates PrototypeJS from core codebase into ingredient, so in future JavaScript library of choice could be used. You may need to manually update your flavours/pub_lib/inc file.

Bug Fixes

• Ensures epadmin upgrade also adds new dataset fields
• Checks whether id_number field is set before parse test in Export::DC and Export::EndNote plugins.
• Fixes data read in hack with Import::BibTeX plugin so it works with unit test.
• Fixes way EPrints::Utils::crypt_password is called in unit tests.
• Fixes validation for date field when it is multiple
• Sets a search_input_style for document license field.
• Fixes is_public check for documents and returns 0 (false) if security field is not set. (Rather than generating a warning, but still returning false).
• Ensure send_out_alert errors if search expression is EPrints::Plugin::Search::Xapianv2 (as well as EPrints::Plugin::Search::Xapian. As such search expressions cannot be used for alerts.
• Fixes bug in defining cookie settings when using /cgi/set_lang>.
• Fixes mismatch in calling code and method (EPrints::Apache::Rewrite and EPrints::Update::Views).
• Various fixes to prevent bad filenames/formats being used or processed.
• Considers datestamp for deleted records when finding the earliest record.
• Updates td element to div in views missed during Accessibility improvements.
• Fixes aria-describedby for non-multiple itemref/dataobjref components.
• Fixes rendering issues for Multipart type field.
• Adds aria-labelledby and aria-describedby to Boolean menu.
• Fixes anomaly with unordered fields.
• Ensures requests with text/* in the Accept header work.
• Moves license/copyright boilerplate for testdata/bin/import_rand_data to ensure it does not break random abstract generation.
• Ensure all fields are added to XML revision files even if export_as_xml is set to 0 (false).
• Fixes journal autocomplete to also group on publisher field.
• Prevents undepositable subjects being unset when eprint is edited.
• Fixes subject and body of test email to stop it getting lost in spam filters.
• Use reply-to rather from to reduce email sent being marked as spam.
• Fixes non-compulsory boolean radiogroup fields cannot deselect UNSPECIFIED.
• Fixes email field validation to not allow commas in domain.
• Fixes date validation so it still allows archive bespoke validation.
• Use default_value for compound sub-field if it is set and value is not.
• Allow uppercase characters in REGEXP_VARNAME to allow for user-defined database password.
• Various minor fixes to bin/check_xapian script.
• Fixes aria-describedby reference when date field used within a compound field. (E.g. in the Dates, Dates, Dates plugin).
• Uses EPrints::Language has_phrase to check whether fieldhelp is set for EPrints::MetaField's get_describedby function rather than generating unnecessary warning messages if missing.
• Fixes DOI import error handling.
• Use CURRENT_USER() to get hostname or database server rather than assuming localhost, when creating MySQL database user and granting privileges.
• Uses more reliable EPrints::Time::iso_date to get current date to use in lift_embargos search.
• Fixes bug with applying mysql_enable_utf8=1 when database driver is not MySQL.
• Sets IsNotNull search condition to 0 rather than empty string for date, int and time type fields.
• Removes invalid characters (x00-x08, x0B, x0C and x0E-x19) from being output as XML, specifically Dublin Core (DC) export.
• Prevents repository admins from modifying usernames where they would create duplicates.
• Display UNSPECIFIED in subject tree if the sort name that would be displayed is an empty string. (This ensures it is still clickable).
• Removes EPrints::Apache::Beast as is no longer needed and has unsupported dependencies.
• Fixes Pod2Wiki to not automatically assign own category to generated API pages (e.g. EPrints::DataObj::EPrint should not have category API:EPrints/DataObj/EPrint).
• Prevents non-ASCII characters causing index tasks to fail when using Xapian.
• Ensures $c->{cookie_domain} can still be set if $c->{host} is not set. (This should make little difference, as no cookie domain means only allow for the current domain).
• Includes EPrints::DataObj::SAX::Handler in use list for EPrints. (Previously present but not in use list).
• Fixed regular expression for matching browse view URLs on Apache::Rewrite.
• Ensures password length validation cannot break if parameters undefined.
• Provides better handling of error if TeX::Encode Perl library is not installed.
• Fixes assumption that PostGreSQL table_schema will always be 'public'.
• Fixes typo on EndNote export to check isbn field is set before exporting it rather than checking issn before trying to export isbn.
• Removes duplicate definition of $c->{make_orderkey_ignore_extras}.
• Removes dc:doi element that does not exist in Dublin Core use dc:identfier dc="dcterms:doi" attribute instead.