eprints / eprints Public

dump_access can now --remove access logs after dumping.

Fixes #11.

According to NYTProf calling {dbh}->{Driver} is really slow, plus DBD::mysql
gives us a utility to enable utf-8 anyway.

00:00:00 is a valid time, so we need to restrict our datetime-range matching
fix to just setting MM-DD fields to 0 in the database.

E.g. if the user supplies '2012' we write 2012-00-00
This will match a search for "..2012-01-01" (as lift_embargoes does)

Fixes #35.

Database::get_ids_by_field_values() should process %opts fixing #32

Fixes #21.

Most of the time lift_embargoes will cause an eprint commit (/revision) due to
fileinfo changing. However if there is no icon on the document the eprint will
not be committed, causing any triggered behaviour (e.g. full_text_status) to
not occur.

This change calls a force-commit on the document (which will force-commit the
eprint in turn).

Otherwise range matching will match the undefined date.

SELECT
  foo
FROM
  bar

Previous bug-fix was a bit over-keen on splitting initials out. We'll now limit
it to a block(s) of upto 3 uppercase letters.

Instead of UNIONing 'Or' conditions, use a logical OR.

Added some more unit tests, but it is possible there is an edge case that
removing the UNION'ed sub-queries was intended to fix. Or it could be that edge
case has been fixed elsewhere (Subjects?) meaning we don't need the sub-queries
anymore.

Fixes #31.

If securecode is NULL (e.g. bad external log-in code) then any user from the
same IP can gain access if they don't have a securecode cookie set.

Fixes #22.

(If no other actor is defined)

If a program has got itself into utf-8 mode errors will spew forth from UUID
generation.

Fixes #25.

Update::Views calls get_record() for the complete result set on a page. This
can break memory/processing limits when generating views on large sites.

As a crude fix we'll apply a limit to the maximum number of items on a single page. It's not clever to send huge HTML files to users anyway.

Export correct subject.depositable value, fixing #29

Support uuid="..." on objects.

Set the .xsd in EPrints XML exports with a pretty /cgi/schema/ep2.xsd.

Fixed subobjects/compound/name missing minOccurs=0.

Don't include sub-fields of compounds/multiparts in schema.

Where Text/Id fields have custom maxlengths specify these individually.

Generate MetaField::Name as a bespoke schema, avoiding repeating it for every
name instance.

Changed schema type names to use '.' instead of '_' (which can be confused with
field names).

Field types with no schema override will just be "xs:string".

Fixup 3.2 custom search export URLs #26