Dear all,
happy Christmas and lots of new features. Have fun!
The file is at ftp://ftp.ebi.ac.uk/pub/software/unix/srs/srs5.1.0.tar.gz
Release notes:
SRS version 5.1.0
=================
-- new features
1) New hierarchical link index allows taxonomic queries and in general
queries on hierarchical entry relations.
2) New named indexes group several index files with the same meaning
into a single "virtual" index.
3) Created databanks of databanks which allows to search which servers
provide specified databanks.
-- changes in the www interface
1) New databank selection page, which uses JavaScript and compatible
with Netscape 4.0 and Microsoft Internet Explorer 4.0 (and the
following minor releases of the programs). Allows working with
large sets of databanks.
2) Following user demands, we added new Query form, similar to the
one present in srs 4.0. Allows simple views to be directly specified
in the query form. The srs5 query form is still available as
"alternate query form".
3) In the query result page the user can now select if the operation to
be performed is on the "selected" entries or on "all but selected".
The second option allows to link or save a set of entries without
need to go to the query manager.
4) The View Manager has been simplified by inserting a new front page
showing the different options from view management. In the list for
selecting the linked databanks, the new named links have been
included
to allow linking to entire databank groups (i.e. searchDB=all
sequence
databanks) with one click.
5) The databank information page now shows if a databank is currently
being reindexed and it can show the progress of indexing.
-- changes in the Icarus language
1) Added objects as a new basic type that can be assigned to variables.
Object have methods. Predefined classes Strv (a string class),
Sequence,
Job.
2) Token tables now accept objects as tokens (using $Wrt:[o:$blabla]).
Very useful for streaming and reorganizing large chunks of text.
3) Added the new parsing concept of TINs (Token INdices). The lazy
parsing
model has beed modified to allow token tables to be dependent from
TINs.
-- changes in the server maintanance
1) Completely changed the way subentries are described, now using TINs.
This modification is NOT backward compatible with the previos
version
of subentries, but because we don't know of anybody specifying
subentries
yet, we hope not to give too much trouble :O)
2) New way of adding applications, makes interfacing applications
even using remote shells (rsh) a breeze.
3) Now it is possible to limit access to internal resources
(databanks, applications or even single fields within a databank)
depending from the internet domain which requests access.
4) The way sequences are treated is changed, and it writes the sequence
in the token table as a "sequence" object. This object has various
methods
which allow sequence conversion in all major formats.
5) All existing parsers have been modified to take advantage of TINs
and of
sequence objects.
6) Inserted the indexing log file as a databank, this allows
to quickly spot indexing problems by querying the log file.
7) SRSCHECK now generates calls to SRSBUILD using the new -nn option
which causes SRSBUILD to be completely silent (no numbers written
during indexing)
8) GETZ has a new option -stat <db_name> which displays the status of
the
indexing process for the specified databank.
The version has been compiled and tested on IRIX5.3, IRIX6.2,
OSF versions 2 and 4, Solaris, Linux. Almost all compiler warnings have
been removed. The ones left are inconsequential.
Thure Etzold, etzold at ebi.ac.uk
Giorgio Verde, verde at ebi.ac.uk
22 Dec, 1997
